APA Newsletter on Philosophy and Computers
Vol. 12, No. 1 (Fall 2012)
Peter Boltuc, Editor

From the Editor
Peter Boltuc, University of Illinois–Springfield

There are many interesting articles in the current issue, though some, including a note from the Chair, didn’t make it in on time. The APA seems to be getting organized, sticking to deadlines quite a bit more than it did in the past, which bodes well, in the long run. We begin with an interesting paper in neurophenomenology, which argues that geometric (platonic) models embedded within some future connectionist architectures may give us the best chance of explaining consciousness. The article, by W. Duch, builds largely on recent work in neuroscience, e.g., by Damasio, as well as on Gärdenfors’ framework of conceptual spaces.

P. Grim and his colleagues bring us a truly delightful paper that highlights philosophical issues in their research on modeling social polarization. This kind of philosophical reflection and analysis of how various assumptions influence the outcome of an investigation is what usually seems lacking in statistical research and modeling, especially in social sciences. The authors show how different notions of polarization (and other not-obvious assumptions) lead to vastly different models and predictions. I consider this approach to be the true applied philosophy of social sciences. Along the lines of computer modeling, S. Ophir presents a graph of philosophical influences among the top philosophers, based on cross-references in Google Books. There are some unexpected outcomes of this search, which pertain especially to the role French Enlightenment used to play—its legacy is so entrenched in our culture that we no longer notice the influence.

The second part of this issue pertains to moral theory. J. Søraker asks about the criteria we want to use in order to evaluate whether the technology enhances well-being. The author proposes a mixture of subjective criteria and objective recommendations. In his eye-opening paper D. E. Wittkower asks the sort of question that seems to have crossed everybody’s mind, "What kind of friends are Facebook friends?" The author claims that traditionally online friendships have lacked the aspect of "hanging out," an essential part of real-life friendship. Mobile technologies take care of this problem, or so it seems. The article reads very smoothly, so that one is not surprised that Wittkower is a best-selling author writing on the issues related to technology and philosophy.

This is the Turing year, and I am sure many readers participated in any number of related conferences organized in the U.K. and elsewhere. In the current issue, R. L. Zebrowski discusses recent approaches to the Turing Test, ranging from Dennett, Harnad, Lackoff, and Johnson, all the way to Brooks’ critique in terms of embodiment. Zebrowski gives the good old Turing Test an interesting defense against this well-presented theoretical background. The last article in this issue features children. I. Oved asked philosophically minded young teenagers, "Where are you, and who are you, when playing a video-game?" The answers may remind you of a debate on the ontological status of web-based objects between L. Baker, A. Thomasson, R. Kurtz, and others, which we hosted in a couple of the recent issues of this Newsletter. We are also glad to have K. Bimbo’s review of A Very Short Introduction to The Computer by D. Ince, published by Oxford. I am glad to mention that three members of the Committee published papers in this issue. I hope the trend continues.


Back to top

Mind-Brain Relations, Geometric Perspective, and Neurophenomenology
Włodzisław Duch, Nicolaus Copernicus University, Poland, and Nanyang Technological University, Singapore

Abstract

This paper presents the rationale for creating geometrical approaches that may help to understand mental processes and relate them to neural processes. Current phenomenology for describing mental processes is not sufficient to create first-person science. Models exploring relations between neurodynamics, geometric representation of mental events, and inner perspective are the best chance for progress in this area.

Understanding the Mind

Two general approaches to mental processes responsible for consciousness and behavior are currently pursued. Psychologists approach the problem with verbal theories based on high-level concepts, such as "intelligence" or "working memory," that are useful to characterize behavior but quite hard to precisely measure and link to neural activity of the brain. Neuroscientists tend to think that more details are needed, and once they are known, cognitive mechanisms will become clear. Michael S. Gazzaniga, a pioneer of cognitive neuroscience, does not share this opinion:

This is what I have been advocating for a long time (Duch 1994-2011), trying to create a new intermediate level of description between neuroscience and psychology. Brain processes described by parameters derived from neuroimaging show activity of individual neurons and neural cell assemblies that are difficult to relate to mental events and inner experience. Psychological description of mental events is largely a confabulation that ignores the real neurodynamical forces responsible for activation of mental states. Consciousness will seem less mysterious if good metaphors can be found to talk about models of mental processes, metaphors that will help us to imagine what really goes on in our minds.

We are visual beings; cortex engaged in analysis of visual information takes a significant part of the brain, and thus the most satisfactory way to understand is to see or imagine. A geometrical model of the mind providing space in which mental events take place, linked with neurodynamics that are responsible for these events, could have a great explanatory power.

Inner Perspective

Detailed simulations of the brain will not necessarily help to understand the mind in a conceptual way. On the other hand, attempts to describe inner experience, or phenomenal consciousness, started in the nineteenth century by introspective psychologists (Fechner, Wundt, Titchener, Würzburg school), and followed by phenomenological movement originated by Husserl, failed to agree on such basic issues as the existence of imageless thought. Phenomenology eventually led to a deep analysis of perception by Merleau-Ponty. Heidegger has focused on embodiment and situated cognition, paying attention to affordances. Varela formulated an experimental phenomenology and neurophenomenology program trying to link the first person approach to the objective science (Varela 1996).

However, recent experiments with random descriptive experience sampling technique (Hurlburt and Schwitzgebel 2007) and Schwitzgebel’s (2010) discussion of phenomenal experience cast serious doubts on the feasibility of science of inner experience. The main reason for this difficulty is rather clear: metaphorically speaking, mind may be seen as a shadow of neurodynamics. Almost all processes that determine our thoughts and behavior are hidden behind the scenes, inaccessible to conscious introspection (Lewicki and Hill 1987). The richness of neurodynamics makes the verbal description of inner experience possible only in a restricted way, using symbols for commonly encountered categories suitable for communication.

Modern Platonic View

In the famous allegory presented in Plato’s Republic, prisoners in a cave are able to see only shadows of real things projected on the wall, while the task of the philosopher is to perceive the true form of things. Mental events resemble such shadows of reality, but instead of idealized forms what we are able to experience can be best described at the level of neurodynamics. Mental events reflect in an active way those features of the environment that are important for survival, and with the modern "mind reading" techniques based on various forms of neuroimaging (EEG, MEG, fMRI, NIRS) our ability to reconstruct mental experience from brain activations is improving (see, for example, Nishimoto et al. 2011). Although many genetic and biochemical processes that change glia and neural cells and their interactions are relevant to understanding behavior, neural correlates of mental experiences are probably captured with sufficient accuracy at a more coarse level of neurodynamics, rapid changes of electrical activations of neural assemblies in the brain.

Biochemical processes responsible for changes in the brain at the molecular level, including neuroplasticity, determine potentially accessible brain states. At this moment mental states that I may experience are determined by the general structure of my brain (evolutionary factors), neural pathways that have been formed as a result of development in infancy followed by life-long learning, and recent priming by experiences (including thoughts and feelings) that changed the landscape of potentially accessible brain activations. All these factors determine the space of my potentially accessible mental states at this moment. Let’s call the current mental state M(B(t)), where B(t) is the brain state at time t and M is a function mapping brain states to mental states.

John Locke (1690) defined consciousness as "the perception of what passes in a man’s own mind." What we are able to perceive are just peaks of neural activity that are sufficiently persistent to be internally categorized, so that the brain may know what is going on. This process requires association of quasi-stable brain activity patterns at a given time with motor activity that allows animals to react, and with activation of symbolic representations (manifested as speech or silent thoughts) in human brains, including representations of self (Damasio 2010).

Mapping continuous activations of the brain B(t) to mental states M(B(t)) leads to information compression, simplifying the decision-making process. Perception needs to be invariant; ripe red apples should look red independent of the illumination conditions. Although information about the spectrum of light reaches the brain, it is not a part of mental experience. Object recognition allows us to recognize functional categories, discover affordances, and prepare for action. Recognizing the mental states of other people serves a similar purpose. The number of categories has to be limited to be useful; therefore, we cannot have an infinite number of symbols to describe precisely all mental states. Although most of the time general categories are sufficient to make decisions, brains may focus on particular experience, experiencing qualia, specific brain activations that may be only roughly categorized at the symbolic level. Strange qualia may be created in unusual situations (Schwitzgebel 2010; Duch 2011), in dreams, hallucinations, visual illusions, peripheral vision, through imagery, strong emotional experiences, or by direct transcranial stimulation of the brain with magnetic field. Brains need to learn through repetition how to categorize new experiences, give them labels, and "explain the experience away" so there is no need to waste time on detailed analysis anymore. Names put the mind at ease, drain neural activation, and relieve anxiety.

Brains and Experience

Most animals have little brains and therefore can distinguish only a relatively small number of ecologically important quasi-stable brain states reflecting their observations and controlling associated actions. Humans with much bigger brains have many parallel competing processes and thus a much higher capacity for diverse mental experiences. Quasi-stable brain states are categorized, associated with other states that win the competition (the Self grabs all the credit, calling this process "I pay attention") and comments are generated: "this is a great wine, with a rich taste, smell, and a deep red color." These comments are intentional, pointing to something in the shared space out there in the world. In fact, they always refer to physical activation of the brain that makes them. They are brain reactions, subjective experiences of qualia that cannot be reduced to discrete categories labeled by a finite set of words. The brain-like systems, by their very construction, have to claim qualia and experiences (Duch 2005).

Brains are truly the meaning machines that physically change themselves. They are changed by top-down causal processes—mental events caused by environment, or by inner activity—and bottom-up processes that may either be initiated by subliminal stimuli or have physiological or genetic origin. We do not have experience with building such systems and analyzing their behavior in conceptual terms. Our conceptual understanding of what brains do is not expressed in terms that are natural from a brain perspective. While the function of the primary sensory cortices may be partially understood using such concepts as spatiotemporal filters for specific information (orientation of the edges, specific frequency, movement, color), we do not seem to have concepts that could be mapped to activation of some brain areas. For example, initially it seemed that area V2 responds to simple visual characteristics (we see what we look for), but now it is known that it responds also to complex shape differences and orientation of illusory contours, and can distinguish between the same stimuli in foreground or background position. Higher visual areas V3-V5 have functions that are associated with some combination of motion, shape, and color, but it may be easier to create a working model than to describe it in conceptual terms. The same is true for other sensory modalities.

Maybe we have not yet learned to pay attention to the kind of information that these intermediate areas are extracting because it may not be that useful in communication. This could be an interesting area of research for neurophenomenology (Varela 1996).

Information in Brains

All signals that reach and change the brain carry some information, although its value may not be immediately accessible at the mental level. Shannon information simply estimates signal entropy and thus is not a good measure of what is intuitively regarded as information relevant for a cognitive system. Calling entropy "information" creates a lot of confusion. The amount of change that is induced in the brain (or in any cognitive system) by incoming signals may be used to quantify information (Duch 2007). Information is not only relative to a cognitive system, but also may have different values depending on its influence on different functions of that system. From a formal Bayesian perspective cognitive system S that has probabilistic model of its environment p(S) is changed by a new observation X to a posterior model p(S|X). The difference between p(S|X) and p(S) measures the size of the change induced by the new observation. It may be understood as the value of the information in observation X relative to the model S. This is not a static quantity; the change in cognitive system may be temporary, it may suddenly restructure the system or slowly grow in time. New information may be generated by the internal flow of brain activations, thinking processes that change the pathways through which neural activation is spread.

The space of potentially accessible mental states evolves over time, reduced due to the brain damage or enhanced due to learning and experiencing new stimuli. These changes may have a profound influence on mental states, but the space of all potentially accessible mental states in the lifetime of a person is always limited. Imagine a space IB={B(L)} of all potentially accessible states that a particular brain may go through in a lifetime L. Individual life history forms a single trajectory B(tL) in this space, describing actual activation states of the brain and resulting in corresponding trajectory of mental experiences. All human brains and their potential states span a space HB of potentially accessible brain states and the corresponding space HM of mental states. Billions of individual trajectories go through this space, but it is still constrained by human brain capabilities, although potentially it may expand enormously due to the brain-machine confluence. The limits of this expansions are not yet fathomable.

The mapping from neural activity to various categories of mental events is slowly elucidated by cognitive neuroscience and it seems feasible that thoughts could be made audible—auditory cortex activation has already been analyzed to reconstruct speech signal (Pasley et al. 2012). Understanding information flow in the brain will be greatly boosted by the Human Connectome project. Models showing how inner, first-person point of view develops in complex systems have not been created yet. In "Self Comes to Mind," Damasio (2010) argues that autonomous systems represented their internal body images and patterns of the environment, reacting emotionally to stimuli that perturb their homeostasis. Self is based on such emotional reactions and in complex brains, memory and feelings, and subjective experiences of emotion, combined with symbolic representations, create conscious, mental perspectives. However, the road from such general understanding to a detailed working model is long.

How can complex processes creating the inner perspective be noted in the overall activity of the brain? I have proposed (Duch 1994-2011) to use visualization based on transformed brain activity signals for bridging the infamous "explanatory gap" between brain and mind, with neuroimaging of neural activity on the physical side and the description of mental events from the phenomenological, first-person perspective on the other (Lutz and Thompson 2003). Visualization should lead to a geometrical model of mental events, showing mind state trajectories that may be linked to neurodynamics, information flow in the brain, and the subjective experience. Mental events, "shadows of neurodynamics," may be seen in low-dimensional feature spaces where each dimension represents a phenomenological property, a specific quality extracted from neural processing.

What kind of phenomenology should be used for such models? It is quite likely that current ways of describing our mental experience will not be sufficient and that we shall have to invent quite new concepts. Description of cognitive systems from this perspective should be more detailed and faithful to the underlying neurodynamics than our current folk psychology allows for.

Language in the Brain

Dynamical system approach to cognition has deep roots in cybernetics. It has been mainly focused on description of external behavior (Kelso 1995), including infants’ development (Thelen and Smith 1994; Smith and Thelen 1994), and sensori-motor activity. In Mind as Motion, edited by Port and van Gelder (1995), other perspectives on the use of dynamical system theory for description of cognitions are introduced, including language.

Flexibility of knowledge representation by patterns of brain activity has not yet been matched by any other knowledge representation framework in artificial intelligence. Jeffrey Elman (Port and van Gelder, Chap. 8) treats representations of concepts as regions in the space of brain activations, and grammatical rules as restrictions on the possible trajectories in this space, leading to the attractor dynamics responsible for language structure. Simple, recurrent neural networks serve him as a model of linguistic systems, predicting the next word in a sentence. The activity of the internal (hidden to the observer) neural units displayed in the principal component coordinates shows the dynamics of this process. Gilles Fauconnier introduced "mental spaces" (Fauconnier 1985) and later conceptual blending (or integration) as a general theory of cognition (Fauconnier and Turner 2002).

Visualization of neural processes during reading shows how associations are made (Dobosz and Duch 2011). Transitions between thoughts are due to neural fatigue, depletion of energy needed to synchronize neural activity (fixing of attention, in psychological terms). Those groups of neurons that are only slightly active may increase their activity and shift the pattern to associated concept, forming a new coalition. Thoughts may jump to a seemingly unrelated subject when a new pattern of neuronal interactions is formed, giving earlier coalition time to refresh. There are many transition pathways between concepts; the dynamics of the process is influenced by noise and many uncontrolled factors, including the history of previous activity. The order of learning matters. Teaching students ancient concepts at the beginning of the curriculum, as it is usually done in philosophy, creates a tendency to evaluate new ideas in terms of the old ones. Mapping from brain activity to events in mental spaces to behaviors and verbalizations is where the "continuity of mind" will reveal itself (Spivey 2007).

Minds are embodied and studies of motor development are certainly relevant for a better understanding of cognition. The ultimate goal of dynamical approach is to describe behavior using differential equations. This, however, is not the same as describing inner experience and in the current form may not be a good candidate approach to bridge the explanatory gap.

Connectionist movement has focused on low-level brain processes and has not been connected with mental events. Paul Churchland (1996) has made the best attempt so far to introduce connectionist ideas to the philosophy of mind, but the influence of these ideas on philosophy and psychology has not been significant. Brainwise: Studies in Neurophilosophy by Patricia Smith Churchland (2002) elucidates many points related to perception, using diagrams that show how perception of male and female faces changes with mouth fullness, nose width, and eye separation. Taste space may be represented using aggregates of sour, salty, and sweet cell responses. The Churchlands’ proposal "that brains develop high-dimensional maps, the internal distance relationships of which correspond to the similarity relationships that constitute the categorical structure of the world" (Churchland and Churchland 2002) follows earlier work of Roger Shepard (1987, 1994) on universal laws and invariants in psychology.

Shepard’s dream to create a theory of mind similar to theories in physics pointed him to search for cognitive universals, natural invariances in perception reflecting properties of the world. Indeed, similarity in brain responses in perceptual and conceptual domains reflects similarity or dissimilarity of observed objects and events in the space of salient features. This space is obviously different for different animal species and different people, changing in many ways with different time scales, depending on the age, current goals, and motivations that control the focus of attention.

Many Faces of Dynamical Cognition

Similar ideas have been floating around in psychology for a long time. Kurt Lewin (1938), recognized as the father of modern social psychology, wrote a book outlining dynamic psychology in 1938 (Lewin 1938). Psychology should represent and derive psychological processes in a conceptual way, but observable facts are not sufficient to achieve it; therefore, hidden "constructs," or "intervening concepts" have to be postulated. Inspired by the successes of physics, Lewin introduced psychological forces. Human behavior should be described by a trajectory in physical space, controlled by forces defined in the hidden, psychological space. Both external and internal forces are responsible for individual behavior and group dynamics.Force Field diagrams, graphs, and tables, introduced by Lewin, are still used in social psychology to analyze factors that push, block, or divert people from pursuing their goals, and determine balance of power. Lewin’s basic idea was that behavioral space should be approximated by discrete meaningful states. Such prototypical states may serve for prediction of behavior, in a similar way as diagnosis of a disease, based on clinical phenotype, serves to select the therapy. Work on probabilistic versions of these ideas continues (Rainio 2009).

George Kelly made another step towards visualization of psychological processes in The Psychology of Personal Constructs (1955), proposing explicit geometrical representation of personalities. In his view, constructs are a relatively small number of psychological processes that can be described using dimensions based on opposite concepts, such as "good-bad" or "happy-sad," useful for making important distinctions. Personal, individual constructs are dimensions of psychological space that characterize people and their mental states, and describe their subjective reality. Results of this analysis are presented in Repertory Grid matrix, with rows representing constructs and columns representing various types of elements, for example, people or mental states (Shaw and Gaines 1992). Personal Construct Psychology (PCP) has been applied to cognitive modeling, and is used in practical applications in social psychology, psychotherapy, personality assessment, and human resources in a business context.

Peter Gärdenfors (2000) introduced conceptual spaces as a framework for modeling representations, using spaces based on qualities (for example, perceptual qualities), phenomenal dimensions that can be inferred from perceived similarities using multidimensional scaling, as Shepard has done. This approach is also aimed at using a high-dimensional geometrical model of cognitive representations for perceptual and linguistic concept representations as an alternative to symbolic and connectionist models. Gärdenfors’ book, Conceptual Spaces: Geometry of Thought, has been very popular and, among many other trends, gave rise to formal conceptual space algebra, the Conceptual Space Markup Language (CSML), an exchange format for sharing conceptual spaces (Raubal and Adams 2010).

Work on knowledge representation in artificial intelligence has brought very similar ideas; for example, Conceptual Knowledge Markup Language developed from the perspective of formal semantics, ontology, and semantic networks. Many applications of conceptual spaces have been developed in such areas as cognitive linguistics, analysis of actions and functional properties of agents, and understanding the dynamics of empirical theories as conceptual change in structured collections of dimensions.

In Continuity of Mind, Spivey (2007) describes mental events as continuous trajectories in the state spaces based on activity of neurons or neural assemblies. This idea follows "neural spaces" that internalize various regularities encountered in the environment (Edelman 2002), for example, tune neurons and sensory organs to respond with maximum discriminatory power to subtle differences found in the environment.

Thinking about temporal dynamics of mental processes, "continuity of mind" moves away from discrete, symbolic representations towards geometric, continuous models. To explain and understand what these models do, Spivey (2007) tries to describe trajectories of the brain activations in a symbolic way using a technique known as "symbolic dynamics," but this is a crude approximation that suffers from combinatorial explosion of the number of symbols necessary to use it in high-dimensional spaces. Moreover, trajectories in the brain-based state space have to be related first to the qualities of experience that are used for description of inner experience.

Mental Models

The field of mental models has been developed with the hope of explaining language and reasoning (Johnson-Laird 1983), with symbolic elements constructed from real and imagined percepts. A large number of possible complex models will of course lead to poor performance. A tendency to focus on a few among many possible models will lead to erroneous conclusions and irrational decisions. Mental models represent explicitly what is true, but not what is false, contributing to systematic errors in thinking. How and why do we reason the way we do? Why should such questions have answers that are independent of neurodynamics? Theory of mental models has ignored many forms of learning, including priming effects that are quite pronounced in experimental psychology and are shown clearly in computational models.

Conceptual constructions used in mental models may be considered true only if they capture simplified neurodynamical processes, but no systematic attempt in this direction has been made. Mental models could probably be derived by analyzing activations of brain subnetworks resulting from priming (or more general learning processes) by the description of the problem, creating probabilistic models of the flow of neural activation, and discretizing this model to express it in a symbolic way. A step towards this has been made using the Fuzzy Symbolic Dynamics (FSD) technique (Dobosz and Duch 2011).

In counterintuitive situations, for example, in the inverse-based rate categorization experiments, psychological conceptualizations are confabulations with insufficient grounding in reality (Duch 1996). If the empirical results were mislabeled, psychological explanations could still be easily created, making this type of explanation not much better than theories of psychoanalysis.

Even simple logical forms of reasoning may be difficult for the brain. Consider the following syllogism:

What precise conclusion follows about the relation between wise men and scientists? Out of 256 possible syllogisms only twenty-four are valid and thus there is ample room for making errors. Indeed, students faced with such syllogism make all kinds of errors and have great difficulty discovering true conclusions. Even when they are told the answer, after a few months they may give the wrong answer during an exam. Syllogisms have been known since ancient times and with proper training in logic errors may be avoided, but numerous natural biases leading to cognitive illusions and errors in thinking have their roots in brain dynamics.

Cognitive Architectures

Cognitive Architectures (Duch et al. 2008; Taatgen and Anderson 2009) are computational models of agents that act and reason using some form of perception. These models are formulated either at the symbolic, conceptual level, using theoretical constructs such as executive controllers, or at the subsymbolic, connectionist, or more detailed neural level where complex cognition should emerge from simpler interactions. In practice many models are based on hybrid combinations of various functions modeled at different levels.

This level of modeling is suitable for extended behaviorism, but it ignores the dynamics of the inner processes. Ultimately cognitive architectures should provide insight into the relation of computations to mental states, although such programs lack persistent dynamical states needed for brain-like information processing (Duch 2005). Bottom-up constraints are easily built in the cognitive architecture models. Top-down factors that should change mental content and lead to a reconstruction of internal models as a result of interaction with environment, or as a result of internal model dynamics, are not yet well captured in such programs. Cognitive architectures have brains that act as control systems, but no minds, no space in which to place dynamic processes where mental content, actual percepts, thoughts, and stream of associations, all information about itself that the system can directly sense. Concepts do not have proper semantics, they are not grounded in perception and affordances; ontologies are a poor substitute for rich imagery of objects and events.

Perhaps a combination of mental models, conceptual spaces, and conceptual blending may be quite fruitful and will help to illustrate why we suffer so easily from cognitive illusions (Pohl 2004).

Geometry of Mind

In geometrical approaches there is no direct relation between objects and events and their abstract representations; only relative similarity relations are preserved. Representation is thus not on the level of individual objects, but rather the whole domains, called by Churchlands (2002) "domain-portrayal semantics."

Although interesting applications of abstract geometrical models may be found (some mentioned in previous sections), the impact of these ideas on understanding cognition has been limited, no links to cognitive neuroscience have been discovered, and very few visualizations that could help in understanding cognitive functions have been made.

Geometrical approaches have some obvious difficulties. First, mathematical similarity functions are symmetric S(A,B)=S(B,A), but psychological similarity, or relations between brain activations representing concepts, are usually not symmetric. In psychology individual preferences are added to context; therefore, similarity from A’s point of view may be distinguished from similarity from B’s point of view. As noted in "Platonic Model of Mind as an Approximation to Neurodynamics" (Duch 1997), the desired properties of similarity measures require the use of Finsler spaces instead of the more familiar Riemannian spaces.

The second difficulty is related to high dimensionality of conceptual spaces. As a result, analysis of even relatively simple, but counterintuitive categorization phenomena, such as the inverse base rate effects, requires at least five dimensions (Duch 1996). Such applications show the danger of conceptual explanations disconnected from neural models that can capture the dynamics of some real cognitive processes: one can invent all kinds of explanations ad hoc, but the true reasons for behavior may be much deeper, and without connecting geometrical models with neurodynamics we will not be able to understand them.

In principle, similarity relations between geometrical representation of mental events should reflect similarities between patterns of activations of the brain areas aroused by these events. Understanding high-dimensional neurodynamical systems described by neural activations seems to be hopelessly difficult unless some way of dimensionality reduction is applied.

Neuroscience provides us with only very rough representation of the brain’s global state, but reading intentions from the EEG signals, used in the Brain-Computer Interfaces, makes constant progress and shows that transformation from brain-related variables to mind-related variables describing intentions is possible, although our conceptual framework will not be sufficient to accurately describe it.

On the technical side there are many approaches to dimensionality reduction that preserve important information, with multidimensional scaling popular for visualization of relations in psychology and latent semantic analysis (Landauer and Dumais 1997) used for visual representations of lexical concepts.

Relevance of Philosophy

Although the brain matter is a necessary substrate for mental processes, detailed neural models may not bring us closer to understanding the mind any more than detailed models of bird wings and feathers would help to understand the dynamics of flight. Cognitive neuroscience is a quite young discipline that tries to create a coherent picture of the brain and its functions. Many research groups are attempting detailed brain simulations, but even if perfect simulation of all functions is achieved, will it help us to understand mental events? The famous Blue Brain project has created precise simulation of cortical minicolumns but it has not proposed any new ideas helpful for conceptual understanding of relations between neural and mental processes.

This is still an area where philosophy, systematic reflection on concepts, can make important contributions, provided that relevant questions are asked. Philosophy has largely ignored reality, staying at the rarefied conceptual level, making itself irrelevant to science. This is true also for many branches of psychology, where theoretical constructs determine the way experts think and analyze experiments. This may be a more subtle extension of folk psychology concepts (Churchland 1996). Like in the joke about the man who was searching for the lost keys near the lantern where there was light, although the keys were lost somewhere else, we search for the knowledge not where it can be found but follow the concepts that structure our ways of thinking.

Neuroscience’s answer to Searl’s Chinese Room (2001) thought experiment is to look at what generates in the brain the feeling of understanding, how the brain may learn proper model of relations between the concept, mapping it to what is already known. One does not need special powers of neurons to do it. Certainly scientists are not going to take such conclusions well.

Some fundamental problems of cognitive science, neuroscience, and brain-inspired cognitive architectures in artificial intelligence may be due to the wrong conceptualization. One such example is the proposal to use kernel methods, popular statistical technique for analysis of data, loosely connected to perceptrons that are rough models of single neurons (Jäkel et al. 2009).

Systematic approximations to neurodynamics may provide an alternative, more fruitful approach. Several questions should be discussed: In view of the complexity of the mind is geometrical approach possible at all? Can approximations to neural dynamics at the mental level be really successful despite limited knowledge of brain mechanisms? Will a geometrical approach indeed help to explain mental processes? Will it lead to a better method of knowledge representation? How should it be approached?

Many (hundreds?) of subspaces connected to specific brain functions may be defined: perception breaks into many components, motor activity memory, social perception, speech—phonemes, prosody, general sounds, emotion, and moral reasoning, abstract concepts. Activity in these subspaces is combined in a combinatorial way in real time due to the competition for attention.

Thus, the final picture of mental processes may be composed of not just one stage with a spotlight of conscious attention, but numerous flashlights that in a coordinated way illuminate important elements on the grand stage of mental space. Sometimes specific lights may be missing and the phenomenological interpretation of the resulting picture becomes difficult or even impossible (Schwitzgebel 2010, Duch 2011).

There is no phenomenology ready to be used for relating the mental scene to existing knowledge. New concepts have to be invented and learned. The final questions are how can we know about ourselves, what may be learned from internal information flow in the brain, and what has to be learned from observation of the results of our own action in the world?

References
Baars, B.J. 1997. In the Theater of Consciousness. Oxford University Press.
Churchland, P.S. and Churchland, P.M. 2002. Neural worlds and real worlds. Nature Reviews Neuroscience 3:903-907.
Churchland, P.S. 2002. Brain-Wise: Studies in Neurophilosophy. Cambridge, MA: MIT Press.
Churchland, P.M. 1996. The Engine of Reason, The Seat of the Soul. Cambridge, MA: MIT Press.
Damasio, A. 2010. Self Comes to Mind: Constructing the Conscious Brain. Pantheon.
Duch, W. 1994. A solution to the fundamental problems of cognitive sciences. International Philosophical Preprint Exchange, all papers are linked to the project page: http://www.is.umk.pl/projects/platonic.html.
Duch, W. 1996. Computational physics of the mind. Computer Physics Communication 97:136-53.
Duch, W. 1996. Categorization, Prototype Theory and Neural Dynamics. Proc. of the 4th Int. Conf. on SoftComputing, Iizuka, Japan, pp. 482-85.
Duch, W. 1997. Platonic model of mind as an approximation to neurodynamics. In Brain-like computing and intelligent information systems, edited by S-i. Amari, N. Kasabov. Springer, chap. 20, pp. 491-512.
Duch, W. 1998. Neural implementation of psychological spaces. Int. Conf. on Neural Network and Brain, Beijing, China, October 1998, pp. 32-35.
Duch, W. 2005. Brain-inspired conscious computing architecture. Journal of Mind and Behavior 26(1-2):1-22.
Duch, W. 2007. Towards comprehensive foundations of computational intelligence. In Challenges for Computational Intelligence, edited by W. Duch and J. Mandziuk. Springer Studies in Computational Intelligence 63:261-316.
Duch, W. 2011. Neurodynamics and the mind. Proc. of the Int. Joint Conf. on Neural Networks, San Jose, CA, IEEE Press, pp. 3227-34.
Duch, W., Oentaryo, R.J., Pasquier, M. 2008. Cognitive architectures: where do we go from here? In Frontiers in Artificial Intelligence and Applications 171, IOS Press, pp. 122-36.
Dobosz, K., and Duch, W. 2011. Visualization for understanding of neurodynamical systems. Cognitive Neurodynamics 5(2):145-60.
Doidge N. 2007. The Brain That Changes Itself. Viking.
Edelman, S. 2002. Neural spaces: a general framework for the understanding of cognition? Behavioral & Brain Sciences 24(4):664-65.
Fauconnier, G. 1985. Mental Spaces: Aspects of Meaning Construction in Natural Language. MIT Press.
Fauconnier, G. and Turner, M. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. Basic Books.
Gazzaniga, M.S. 2010. Neuroscience and the correct level of explanation for understanding mind. Trends in Cognitive Sciences 14(7):297-92.
Gärdenfors, P. 2000. Conceptual Spaces: The Geometry of Thought. MIT Press.
Hurlburt, R. and Schwitzgebel, E. 2007. Describing Inner Experience? Proponent Meets Skeptic. MIT Press.
Johnson-Laird, P. 1983. Mental Models: Towards a Cognitive Science of Language, Inference and Consciousness. Harvard University Press.
Jäkel, F., Schölkopf, B. and Wichmann, F.A. 2009. Does cognitive science need kernels? Trends in Cognitive Sciences 13(9):381-88.
Kelly, G. 1955. The Psychology of Personal Constructs. New York: Norton.
Kelso, J.A.S. 1995. Dynamic Patterns. Bradford Book, MIT Press.
Landauer, T. and Dumais, S. 1997. A solution to Plato’s problem: the latent semantic analysis theory of acquisition, induction and representation of knowledge. Psychological Review 104:211-40.
P. Lewicki and T. Hill. 1987. Unconscious processes as explanations of behavior in cognitive, personality, and social psychology. Personality and Social Psychology Bulletin 13:355-62.
Lewin, K. 1938. The Conceptual Representation and the Measurement of Psychological Forces. Durham, NC: Duke University Press.
Lutz, A. and Thompson, E. 2003. Neurophenomenology: Integrating subjective experience and brain dynamics in the neuroscience of consciousness. Journal of Consciousness Studies 10:31-52.
Nishimoto, S., Vu, A.T., Naselaris, T., Benjamini, Y., Yu, B., and Gallant, J.L. 2011. Reconstructing visual experiences from brain activity evoked by natural movies. Current Biology 21:1641-46.
Pasley B.N. et al. 2012. Reconstructing speech from human auditory cortex. PLOS Biology 10(1):e1001251, 1-13.
Pohl, R. 2004. Cognitive Illusions: A Handbook on Fallacies and Biases in Thinking, Judgement and Memory. Psychology Press.
Port, R. and van Gelder, T.J., eds. 1995. Mind as Motion. MIT Press.
Rainio, K. 2009. Kurt Lewin’s dynamical psychology revisited and revised. Dynamic Psychology 2009:1-20.
Raubal, M. and Adams, B. 2010. The semantic web needs more cognition. Semantic Web Journal 1(1-2):69-74.
Schwitzgebel, E. 2011. Perplexities of Consciousness. MIT Press.
Searle, J. 1980. Minds, brains and programs. Behavioral and Brain Sciences 3:417-57.
Shaw, M. and Gaines, B. 1992. Kelly’s ‘geometry of psychological space’ and its significance for cognitive modeling. The New Psychologist 10:23-31.
Shepard, R. 1987. Toward a universal law of generalization for psychological science. Science 237:1317-23.
Shepard, R. 1994. Perceptual-cognitive universals as reflections of the world. Psychonomic Bulletin & Review 1:2-28.
Smith, L.B. and Thelen, E., eds. 1994. A Dynamic Systems Approach to the Development. MIT Press.
Spivey, M. 2007. The Continuity of Mind. New York: Oxford University Press.
Taatgen, N. and Anderson, J.R. 2009. The past, present, and future of cognitive architectures. Topics in Cognitive Science 1:1-12.
Thelen, E. and Smith, L. 1994. A Dynamic Systems Approach to the Development of Cognition and Action. MIT Press.
Varela, F. 1996. Neurophenomenology: a methodological remedy for the hard problem. Journal of Consciousness Studies 3:330-49.


Back to top

Philosophical Analysis in Modeling Polarization: Notes from a Work in Progress
Patrick Grim, Aaron Bramson, Daniel J. Singer, Steven Fisher, Carissa Flocken, and William Berger, University of Michigan

Introduction

Computational modeling and computer simulation have quickly established themselves not merely as useful add-ons but as core tools across the range of the sciences. We consider computational modeling to be a promising approach to a range of philosophical questions as well, and to questions that sit on the border between philosophy and other disciplines (Burkholder 1992; Bynum and Moor 1998; Holyoak and Thagard 1997; Grim, Mar, and St. Denis 1998; Grim 2004). Questions regarding the transference of belief, social networks, and opinion polarization fall in the latter category, bridging epistemology, social philosophy, sociology, political science, network studies, and complex systems. This is the focus of some of our current research.

Our purpose here is not to sing the praises of computational modeling as a new philosophical technique. Our purpose is rather to emphasize the continuity of computational model-building with the long philosophical tradition of conceptual analysis (Hanna 2000; Sandin 2006; Beaney 2009). With reflections from the process of building a specific model, we want to emphasize two points: that (1) the work of constructing a computational model can serve the philosophical ends of conceptual understanding, in part because (2) attempts at computational modeling often require clarification of the core concepts at issue.

I. Computational Modeling and Philosophical Analysis

In their final form, papers in scientific computational modeling always look perfect: they appear to be the work of a rational investigator who thought things through step by step in advance: from methods, to results, to discussion and conclusion. It’s good that these papers look that way—good for brevity, evaluation, and use in future work. That is how we want our work on belief networks and polarization to look eventually.

But, of course, the polished published form of a paper can give an entirely misleading impression of the research trajectory—the impression that both the conceptual work at issue and the path of design and programming were neat, tidy, and fore-ordained. Almost inevitably, they were not. We will use our current work in progress as an example. Here, unlike its future final form, we will lay out the research in something more like real time, complete with fits, starts, and second thoughts. A key point is that those fits, starts, and second thoughts often indicate the need for philosophical analysis in a fully traditional sense. Computational modeling calls for and enforces a full and explicit conceptual understanding of what it is one is trying to model. To employ computational techniques, one must have a full and explicit understanding of what it is one is trying to find out, within what parameters, with what background assumptions, and why (Pollock 1998).

We offer our current work on belief polarization as a case in point. The history of this project is one in which we have repeatedly had to ask what abstract representations of social information contact were plausible. We’ve had to ask and ask again whether certain modeling assumptions were realistic portrayals of belief and trust, and whether it matters the extent to which they clearly were not. The history of the project is one in which we have repeatedly had to return to questions of how to define and measure the phenomenon we were after, and even whether there was just one phenomenon at issue. This exploration, which is at the edge of various sciences, has repeatedly demanded far more than computational resources. Flying under the colors of updating algorithm design and definition of quantitative measures, for example, we repeatedly found ourselves doing just good old-fashioned conceptual analysis in a new-fangled computational terminology.

II. Understanding Polarization: Initial Motivations

What we wanted to know about was polarization of beliefs in society. We started with the impression that the increased polarization of America was an agreed and established sociological fact. Everybody talks about it and a range of books are written about it (McCarthy, Poole, and Rosenthal 2006; Brownstein 2007; Hetherington and Weiler 2009; Fiorina, Abrams, and Pope 2010), so we thought it must be real.

The idea was to use the tools of agent-based modeling to try to understand that polarization better—to understand the factors that influence polarization: factors necessary for polarization, perhaps a handful of factors sufficient for polarization, and perhaps even social measures that could be used to reduce polarization.

At the beginning, we had a hunch that increased polarization in America might have something to do with the structure of media sources. The core idea was the following: We seem to have been less polarized when there was essentially one source from which everyone got their news: the Evening News on ABC, NBC, and CBS. The news coverage on the three major networks was essentially interchangeable—all a version of Walter Cronkite. All followed a journalistic code that insisted that editorializing be kept strictly separate from reporting.

News is no longer like that. Fox News and MSNBC have obvious political slants, are positioned at rival ends of the political spectrum, and do not seem to care where journalism leaves off and the editorial begins. Perhaps the change in where we get our news has something to do with why America is so polarized.

That was the initial motivating hunch. Could a model illustrate those belief dynamics? Could it show us whether split news media was an easy route to polarization, or even a possible route? Could it give us hints as to what kinds of factors might ameliorate or reduce polarization?

We had worked previously with networks of artificial agents whose beliefs were modeled as numbers between 0 and 1 and who updated those beliefs in terms of the other agents with whom they had contact. We had used that abstraction in the context of investigating infection, belief transference, and genetic crossover as alternative modes of information diffusion on networks. All that work saw final presentation in polished form (Grim, Reade, Singer, Fisher, and Majewicz 2010; Grim, Singer, Reade, and Fisher 2011; Grim, Singer, Reade, and Fisher 2012).

Figure 1. Types of linked sub-networks used in previous work on belief and infection dynamics (Grim, Reade, Singer, Fisher, and Majewicz 2010; Grim, Singer, Reade, and Fisher 2011).

Figure 2. Histograms and networks constructed to match degree distributions drawn from data within the Black and White communities, Pittsburgh Random Household Health Survey (Grim, Thomas, Fisher, Reade, Singer, Garza, Fryer, and Chatman 2012a, 2012b).

Figure 3. Trust levels in the Black community correlated with network position, Pittsburgh Random Household Health Survey. Red nodes indicate low trust; blue nodes indicate high trust (Grim, Thomas, Fisher, Reade, Singer, Garza, Fryer, and Chatman 2012a, 2012b).

Figure 4a. and 4b. Two ways of graphing trust updating. In each case an agent increases trust as shown in an agent with a belief less than  in distance from his own, and decreases trust as shown in an agent with a belief greater than τ from his.

Figure 5. Horizontal location represents belief. Snapshots show a typical evolution of random network with global trust updating. Generations 5, 15, 25, and 30 shown.

Figure 6. Horizontal location represents belief. Snapshots show a typical evolution of random network with local trust updating. Generations 5, 15, and 30 shown.

Figure 7. Attitudes toward abortion, distribution by year, from the full sample General Social Survey 1997-1994 (DiMaggio, Evans, and Bryson 1996, p. 709).

Polarization type 4: Size disparity

A society that has one dominant opinion group with a few small minority outliers seems less polarized than one with a small number of comparably sized competing groups. Groups are more polarized in this sense if the different beliefs are held by equal numbers of people. Using the notation that G is the set of groups, and _i is the size of group i, size disparities can be measured by calculating the absolute deviation: 1/(2N) ×  | _i - µ_G |. This is just the normalized sum of distances from the mean community size; it equals zero when all the groups are the same size and increases the more groups differ from the mean size. It maxes out at 1 as the number of groups and size differences go to infinity, making it a nice measure for comparison across different configurations.

Views on women’s role in public life are no longer as polarized in this sense as they once were, even though there are small groups who continue to hold anti-feminist views that were once much more common. In the past, major portions of the population fought racial integration vociferously. Even if the views represented there are still held by some, polarization on the issue of racial integration has clearly decreased.

Polarization type 5: Coverage

We think of polarized societies as having a few tightly packed sets of beliefs. The inverse of this, a broad spectrum of beliefs, can be captured in a variety of ways. One example is the proportion of the belief spectrum held by members of society. The larger the areas of unoccupied belief space, the more polarized the society. The more focused and less diverse the beliefs in a society are, the more polarized it is.

A simple way to envisage the measure in a discrete instantiation is to think of the spectrum of possible beliefs between 0 and 1 as divided into small bins of size d (e.g., d = 0.01 or normalized by setting d to 1 / the number of agents). We can then measure coverage in terms of the proportion of bins filled. Alternatively, we might want a continuous measure over the belief space. This can be done by summing the amount covered by d-diameter halos around each agent; i.e., any portion of the belief space that is within d of an agent is considered covered; the rest is uncovered.

Polarization in the sense of coverage is related to dispersion, but does not include the shape of the belief dispersion. We might therefore think of coverage as a sub-measure of global dispersion, measuring how much dispersion there is without measuring its location.

Polarization type 6: Regionalization

While polarization in the sense of coverage represents how much belief dispersion there is without accounting for where beliefs are dispersed, we might also want to measure certain aspects of belief regionalization without attending to the belief area covered over all. In considering small bins of possible belief, for example, we might mean by polarization not how few bins are filled but the extent to which there are regions of empty bins between regions of bins that are occupied.

With 100 bins, for example, there might be three different cases: (a) that in which bins 1-50 are the only bins filled, (b) the situation in which bins 1-25 and 31-55 are filled, and (c) the situation in which 5-bin regions are filled, separated by 5-bin holes: regions 1-5, 11-15, 21-25, 31-35. . . are the only ones filled. Each of these will be equally polarized in the sense of polarization as coverage. Counting the number of empty regions between regions of occupied spaces, however, gives us a measure of polarization in which (c) is more polarized than (b), which is in turn more polarized than (a). Regionalization seems a further intuitive sense of polarization well worth quantifying.

It should be noted that regionalization per se does not distinguish between the case in which (b) bins 1-25 and 31-55 are filled, and (d) that in which 1-25 and 75-100 are filled. In terms of regionalization that may be exactly what we want: beliefs in the two cases are regionalized in precisely the same sense, though the groups are farther apart in the sense of spread.

Senses 1 through 6 of polarization can all be seen in terms of histograms of beliefs on a single issue across a population. But there are other senses of the term that are essentially (a) multiple-opinion or (b) network-based.

Polarization type 7: Multiple opinion convergence

Given polarized groups on issue A, are these same groups polarized on B, C, and D? The more interlocked rival beliefs are within rival groups, the greater the polarization across the community. Fiorina and Abrams (2008) note that intra-group polarization in this sense may increase even though population distributions on particular issues may not change. Bishop notes that individuals may move to "neighborhoods where others have similar political views, changing their partisan identifications to match their ideological and issue positions" (2008, 578).

Polarization type 8: Community fracturing

Sub-communities may be polarized simply in the sense that there is little or no communication between them. Even if two separated communities have identical and uniform beliefs, that uniformity may be coincidental and temporary.

In Ethnic Conflict and Civil Life, Varshney (2002) demonstrates how group interactions ameliorate levels of inter-group violence, and conversely, how group isolation increases the likelihood of violence. Varshney’s central claim is that "pre-existing local networks of civic engagement between two communities stand out as the single most important proximate cause" for the difference between peace and violence (9). Put another way, cities with social networks that connected Hindus and Muslims through the same institutions were much less likely to see outbreaks of ethnic violence than cities in which Hindus and Muslims belonged to distinct civic institutions.

VII. First Results and Work in Progress

We think we have made progress, along the lines above, in the conceptual foundations necessary to model building with an eye to understanding polarization. Simple assumptions of a single belief scale and belief updating will remain, but with a range of variability to be explored in (a) trust updating functions with (b) different  values against (c) local and global scales, with further variations in (d) social network structures and sizes, (e) initial configurations, and (f) media sources and effects. Our measures in exploring variations in those parameters will be measures of polarization in the distinct conceptual senses outlined above.

The following is a sample of the kinds of results we are headed for.

Begin with a random network of 50 agents, initially assigned beliefs between 0 and 1. Begin with a simple linear function for belief updating. That function is "tune-able": it may be when a contact is within .2 of an agent’s belief that his trust in that contact increases, and beyond .2 that he begins to discount input from that source. Or that -point may be wider: it may be a distance of .3 that marks the difference, or any other number.

Consider now two variations. In one, the -point is marked on a scale calibrated to the entire spread of beliefs across the population. In that case the belief spread of my particular contacts may not be as important. Relative to the range of opinions across the population, all of my friends may think pretty much like me. We will have a mutual opinion admiration society, increasing trust in each other and influence on each other based on trust. This first variation is a "global" updating model. I tend to trust individuals with beliefs like mine, gauged against the whole spread of public opinion.

Consider a second variation that differs only in the scale on which trust updating is measured. In this case -points .2, .3, .4 aren’t measures across the whole spread of beliefs within the population at large. They are measures across just the spread of beliefs of my immediate contacts. In this case it will be guaranteed that one of my contacts is the farthest out—and I will decrease trust in that individual no matter how close our beliefs on the "objective" scale of the entire spread within the population. This is a "local" updating model. I trust those among my contacts with beliefs like mine, gauged against the field of opinion among those with whom I am in contact.

Given the other particulars of the model—a random network of 50 agents and a linear updating function—that difference between global and local scaling makes a major difference in the emergence of polarization, in several senses.

Figure 8 shows a sample of what happens with a  point of .25 and global updating. Figure 9 shows by contrast what happens with a  point of .25 and local updating. More complete animations for each are available at www.pgrim.org/workinprogress.

Figure 8. Horizontal location represents belief. Representative slides from evolution of a random array with a  point of .25 and global updating. Agents update trust positively in those closest to their beliefs, update trust negatively in those farthest away, with a transition point from positive to negative update at t = .25. See also www.pgrim.org/workinprogress.

Figure 9. Horizontal location represents belief. Representative slides from evolution of a random array with a  point of .25 and local updating. Agents update trust positively in those closest to their beliefs, update trust negatively in those farthest away, with a transition point from positive to negative update at t = .25. See also www.pgrim.org/workinprogress.

Figure 10. Results for global (left) and local (right) scaling with the same trust updating function (as in Figure 4a) and different  points from .05 to .75, using the same initial random seed throughout.

Automatic Generation of Philosophers Network from Google Books Repository
Shai Ophir, Starhome, Innovation in Motion

Abstract

The ability to search through millions of books, aggregate the results, and analyze them all together enables a search for new patterns of historical events that could not be identified before. Analytic tools which track the data traffic on the Internet can be used in historical research to identify historical trends. The possibility of having a very large amount of historical texts, images, and other media types under a unified access, search, and analytic engine enables a more complete and balanced view of the historical scene, either on the micro history or the macro history level. This analysis is relevant especially for the history of philosophy, where ideas cross the boundaries of time and location, creating chains of concepts.

A new domain, Digital Humanities, has been established to achieve these goals, and others. However, text analysis projects performed so far were usually based on an import of e-Text corpora into a software-specific internal database. This restriction imposed a major limitation on the order of magnitude of the input being analyzed. This paper reflects an attempt to derive historical knowledge from a very large online corpora of text, Google Books, without the need to import the data to a software-specific storage database. The automatic derivation of a philosophers map will serve as an example. The map will include most of the major philosophers and will reflect the number of references they made for each other as a possible measure for a level of intellectual influence.

Keywords

Philosophers Map, Philosophers Network, Conceptual Map, Social Network Analysis, Citation Analysis

History Analytics: A New Dimension for Research

The online corpora of historical e-Texts may create a tremendous opportunity for a new type of historical research. The ability to search through millions of books, aggregate the results, and analyze them all together may bring to light new patterns of relations and events that could not be identified before. Conceptual changes, for example, may be tracked down, using relations between terms and concepts over the years. It may be possible to identify the exact timing of minor cultural changes that finally brought about major conceptual changes. The following will demonstrate only a small portion of the potential of e-Text research.

The possibility of having a very large amount of historical texts, images, and other media types under one access, search, and analytic engine allows for a more balanced view of the historical scene, either on the micro history or the macro history level. Up until now, the researcher was limited by the manual capacity of searching through hard copy books and a limited set of online texts. Very few historians, such as Toynbee or Spengler, for example, or world-system social science researchers, such as Fernand Braudel or Immanuel Wallerstein, could have a claim for a global macro-history view. The information age may radically change this situation, enabling more sociologists, historians, and scholars to have a broader world view of history. We may discover that significant resources were left outside of the research borders because they were simply unknown to contemporary researchers. Using analytic techniques we may identify "dark matter" of knowledge which was used in the past but abandoned over the years.

Background: Citation Analysis and Science Mapping

Mapping science activities according to reference analysis is an area of research that has been emerging for a few decades already.

The CiteSpace software created by Chaomei Chen demonstrates remarkable analysis of author relations, derived from a corpus of scientific journals published during a specific range of years. Using CiteSpace, pivotal points in co-citation are identified according to their "betweenness centrality," i.e., a factor derived from the number of links connected to many other nodes (Chen 2006, 2007). The software is open for public use on Chen’s website at http://cluster.cis.drexel.edu/~cchen/citespace.

Loet Leydersdorff argues that betweenness centrality is an indicator for transformative changes (Leydersdorff 2007). The argument is demonstrated using co-citation analysis of scientific journals. Shibata further argues for the predictive function of betweenness centrality, where predicting changes in areas of research according to current changes in citation references (Shibata, Kajikawa, and Matsushima 2007).

Other research concentrates on burst detection of references as a measure of a conceptual change (Kleinberg 2002). Garfield developed a practice to predict Nobel Prize winners based on citation analysis (Garfield 1992).

These efforts, as well as many others, show the relevance of citation analysis for the research of the history of science, science development, and the history of concepts and ideas in general. This project adopts the citation reference analysis to the area of philosophy, with major fundamental changes: (1) the references are taken from books, reflecting the 2,500 years of the history of philosophy, and (2) the books are taken from the Google Books online open repository, using a different technique than that used by the citation reference projects so far.

Creating a Philosophers Map: Data and Software Tools

Text analysis projects performed in the past were usually based on an import of e-Text corpora into a specific software tool internal database. The texts had to be fetched, stored, organized, cleaned, and sometimes sorted and managed manually, where inappropriate texts needed to be removed. This restriction imposed a severe limitation on the historical value of the analysis, due to the relatively small amount of text. This paper reflects an attempt to derive historical knowledge from a very large online corpora of text, Google Books, without the need to import the data to any specific internal storage tool.

This paper describes the creation of a philosopher’s network from the writings of the great philosophers of all times. The list is taken from the EpistemeLinks website, a well-known site in the philosophy community, which provides a list of the 75 most common philosophers according to the ranking of their popularity (through access to their writings). The full list of philosophers used for this paper can be found at http://www.epistemelinks.com/Main/MainPers.aspx.

The aim of the software developed for this project is to count the number of references of each one of the philosophers (writings) to each one of the other philosophers on the list. Using such a matrix of references, a network of "influences" or "philosophy concepts" can be created. There are 75 philosophers on the list; hence, the matrix of mutual references is composed of 75x75 relations = 5,625. There is a need, therefore, to search Google Books 5,625 times, for each pair of philosophers, while one of them is the search term and the second is the author. However, there is no viable tool provided by Google for searching Google Books in batch mode. Google Books provides a very preliminary Application Programming Interface (API), which is intended mainly for including links to Google Books from other websites. There was a need, therefore, to develop an automatic search tool that was able to receive a table of search items, invoke Google Books search for each line, and fill in the table with the search results. Google limits automatic search for ~200 sequential queries, and also limits the number of queries per day, so there was a need to split the search into multiple sub-tasks and perform them over a few days.

The matrix of references was converted to a graph of relations by the software package Pajek. This is a popular data visualization tool for network analysis, which is provided for free. More details about Pajek can be found at http://vlado.fmf.uni-lj.si/pub/networks/pajek/.

Another visualization tool used in this project was Microsoft NodeXL, an Excel plug-in dealing with node graphs.

Philosophers Network Visualization and Findings

Figure 1 is composed of the references of the great philosophers to other great philosophers (all within the same group of 75 philosophers). The nodes stand for the philosophers, where the edges (the lines between the nodes) represent the number of references. The thickness of the edges is proportional to the number of references. The network is visualized using the Pajek software tool.

Figure 1. Great Philosophers References Map.

Figure 2. Great Philosophers Central Closeness via Spiral Graph.

Figure 3. Great Philosophers Central Closeness via Spiral Graph – Zoom In.

Prudential-Empirical Ethics of Technology (PEET) – An Early Outline
Johnny Hartz Søraker, University of Twente

1. Introduction

The pace of technological innovation and its prevalence in modern life leaves little reason to doubt that technology has had a profound impact on our well-being. Computers in the workplace reduce the number of challenging activities that require skill, which turns working life into nothing but a means to other ends. The same goes for technologies that allow us to passively absorb entertainment, which for many people constitute the default option when returning from an apathetic or anxiety-ridden day at work (Csikszentmihalyi 1991, 69). With both work and leisure activities being increasingly intertwined with (computer) technology and our volitional activities being the only way to transcend the well-being set-point that our genetic dispositions and circumstances determine (Lyubomirsky, Sheldon, and Schkade 2005), one way to better understand and potentially improve people’s well-being in both domains is to investigate how and when technology tends to improve or worsen our well-being. In this outline I will present an approach, currently in its early stages of development, which analyzes technologies according to their positive or negative effects on well-being, and grounds these prudential considerations in empirical research—while at the same time subjecting this evaluation to ethical scrutiny. To clarify what this involves, I will proceed by explaining each of the four elements of the approach ("prudential," "empirical," "ethics," and "technology") before pulling the strings together and discussing the pros, cons, and limitations of this approach.

2. Prudential...

The realization that technology can drastically change our lives even if there is no wrongdoing involved has led to what we may term an "axiological turn" in ethics of technology (Brey 2006; Brey, Briggle, and Spence, forthcoming; Higgs, Light, and Strong 2000; Søraker 2010). This is a natural consequence of the realization that technologies can change our lives radically without there being any right- or wrongdoing involved. Few technologies have changed our daily lives as much as the television, but it seems pointless to try to point fingers at someone to be blamed for this effect. This means that when evaluating technologies where no direct wrongdoing is involved, the traditional way of applying ethical theories also becomes insufficient. As a complement to considering the ethical, societal, and political implications of technology, the axiological turn asks us to also consider the prudential effects.

The term "prudential" (or "prudential value") refers to something that is valuable for someone, contrasted with something that may be good in itself (if there is such a thing) or something that is good for something (which would typically be an instrumental value). To take an example, the Mona Lisa may have a certain aesthetic value in itself, it may have a certain instrumental value in virtue of its value for something else (e.g., for the Musée du Louvre, for the French republic, for the Italian sense of pride, etc.)—but it also has a certain prudential value in virtue of being good for some individual human beings. If something is good for you without being merely instrumentally good, then we are talking about how that thing contributes to your well-being. This is the sense in which I use the term "prudential" for present purposes.

This immediately raises the question of what it really means for something to be good for someone. This is a long-standing debate that cannot be fully accounted for within the scope of this paper, but one of the main dividing lines lies between objective and subjective accounts. Briefly put, objective accounts of the good life prescribe one or more "goods" that must be present for someone to genuinely have a good life—such as the necessity of friendship as expressed in Aristotle’s Nicomachian Ethics (book VIII, 1). Unlike subjectivist accounts, objectivist theories of the good life can make the claim that one who experiences one’s life as good may be completely wrong about this, and that the good life does not have to be experienced as such. Subjectivist accounts, on the other hand, prioritizes first-person, subjective experience; thus, a life that is experienced as good is ipso facto a good life (although there will typically be a few additional conditions that have to be met, such as being sufficiently informed, capable, and rational).


The advantage to objective accounts is that they can more substantially claim that one has deluded oneself into an existence that may appear to be a good life, but which in reality cannot be a good life—thereby accounting for the ways in which delusion, confabulation, and various forms of mental illness may hinder us from understanding what is good for us. The advantage of the subjective accounts lies in taking the individual seriously, and allowing for a plurality of ways in which to attain well-being. As Griffin aptly puts it: "To get an account of well-being that would be of use in moral theory we have to move . . . on to what is valuable to the particular person affected in each case we judge"(Griffin 1998, 72). I have argued this point at length elsewhere (Søraker 2010), but there is a need to find a golden mean between these positions—a golden mean between staunch paternalism (you are wrong about what is good for you) and complete relativism (anything goes). I think we can find such a golden mean in the notion of "recommendations," which may be of the form: This tends to increase the subjective well-being for most people, so it might be worth trying. Admittedly, this leaves the notion of well-being a contingent one, but we can still add substance to it by grounding these recommendations in empirical research (in addition to ethical considerations, which I will return to later).

3. ... Empirical ...

Within the aforementioned axiological turn, a technology is typically assessed according to its agreement with a philosophical theory of the good life. For instance, it has been argued that relationships in virtual worlds cannot meet the Aristotelian criteria for being the kind of relationship that contributes to the good life (Fröding and Peterson, forthcoming), and that technology-driven consumerism threatens non-material and non-hedonistic values (Brey 2007). Although these analyses may be instructive, there is often a lack of attention to whether the applied theories are actually true. If a technology fails to meet certain philosophical conditions for a good life, it is typically left as an open question whether those philosophical conditions in reality lead to better lives—was Aristotle really correct when claiming that genuine friendship is necessary for a good life, and do material and hedonistic values really make us unhappier? Together with numerous other problems with applying philosophical theories of the good life to technology assessment—including naturalistic fallacies, intuition pumps, and imprecise terminology¹—a more substantial way to ground the analysis is to consider empirical research on what actually tends to increase or decrease our well-being. This does presuppose a subjectivist account of well-being, since this empirical research is typically based on self-reports, but it allows us to provide recommendations grounded in concrete activities that have a strong tendency to increase most people’s well-being (more on this later).

In his 1998 inaugural address, former president of the American Psychology Association Michael Seligman argued that psychology should not only be a science of negative mental health—explaining, predicting, and curing mental illness—but also a science of positive mental health: explaining, predicting, and enabling well-being. Seligman thereby founded "positive psychology," which is the empirical study of the kinds of events and experiences that tend to lead to well-being and what can be done to improve it. This field has grown tremendously since then, with its own conferences, journals, and graduate programs throughout the world.

Although controversial and fraught with methodological and conceptual problems, one of positive psychology’s advantages is that it pays tremendous attention to these problems, resulting in a level of self-scrutiny unparalleled by many other empirical disciplines (cf. Lopez and Snyder 2003; Ong and Van Dulmen 2007; Peterson 2006). Although findings are typically based on self-reports, whether in real-time ("What are you doing right now, and what is your level of well-being?") or in retrospect ("Overall, how satisfied are you with your life?"), increased validity is sought by means of extensive meta-studies, correlations with other statistics, neuroscientific and evolutionary explanations, as well as trying to control for cognitive biases. At the end of the day, "subjective well-being" remains an operational term, and it can be questioned whether this actually corresponds to a deep philosophical notion of the good life. Nevertheless, it is the most concrete, substantial, a posteriori account of which events and activities that tend to make people happy we have, and the methodological challenges should simply be a reminder not to apply these findings blindly.

One discovery is that we have a "set point" which partly determines how happy we can be, but an equally consistent finding is that there are still numerous "volitional activities" that can bring us up to or beyond our set-point of well-being (cf. Lyubomirsky et al. 2005; Peterson 2006). After some ten years of intensive research, a scientific picture of human well-being is starting to emerge, constituted by a large body of empirical findings that has largely been left untapped in ethics of technology. Many of these findings can be "translated" to concrete technological features, as I will return to below, but we cannot do so with the purpose of achieving a quick fix of happiness. Given the methodological challenges and the somewhat controversial practice of quantifying well-being, we need to bring ethics into the approach.

4. ... Ethics ...

Rather than drawing the notion of well-being from philosophy, as most other similar approaches do, PEET draws the responsible application of these findings from philosophy. As I will illustrate by way of examples below, most of the events and experiences that have been shown to increase well-being can often come with negative effects as well, either due to having multiple effects on well-being or due to introducing side effects that are undesirable for other ethical, cultural, or political reasons. Most fundamentally, something being good for me may, of course, be impermissible because it is not good for others. If we design technology solely according to what is directly beneficial for the users’ well-being, we might lose sight of any negative effects the same technology may have—not only on the users (which I will return to in the next section) but also unintended side-effects on other users. For instance, it has repeatedly been shown that acts of kindness increase one’s own well-being (Lyubomirsky et al. 2005; Otake, Shimai, Tanaka-Matsumi, Otsui, and Fredrickson 2006), but since positive psychology is primarily concerned with subjective experiences, the same effect can be achieved from the mere illusion that one does acts of kindness. To illustrate, we can imagine an immersive virtual environment where simulated acts of kindness give rise to increased well-being, when those acts in reality have no effect on the actual world at all. Indeed, "liking" a cause or charity on Facebook may give us such an illusory feeling that we are being altruistic, when the action in reality has little if any effect. This could provide the user with increased well-being, but it might make us less inclined to act similarly in the actual world. Likewise, well-being gained from being social in virtual worlds could leave us with a society where there is less and less need to meet others in the flesh, with all the ethical, political, and cultural ramifications that may have.

This may sound like a purely utilitarian deliberation, weighing the positive consequences for myself against the negative consequences for others, but this is where it is important to distinguish between prudential and ethical value. Separating between prudential and ethical value allows us to operate with subjective criteria for the former, while including objective criteria for the latter. For instance, we can invoke the categorical imperative on the ethics side, and disallow an activity that entails using a person as a mere means towards this end, regardless of how beneficial it may be for subjective well-being. The prudential considerations can also be complemented with virtue ethics, for instance, by arguing that a flow-inducing activity is in disagreement with what it means to be a virtuous person. In short, even if PEET emphasizes the importance of applying empirical findings in a responsible manner, it leaves it open which ethical theory to "plug in." A utilitarian can simply enrich her utilitarian calculus with empirical findings, a Kantian can use empirical research to evaluate (only) that which is morally permissible, and a virtue ethicist can forbid any beneficial activity that is inconsistent with what it means to be a virtuous person. Furthermore, the analysis can (and should) be augmented by political and cultural considerations, such as whether a prudentially good activity could be detrimental to democratic ideals, lead to social unrest, or threaten cultural norms and values.

5. ... of Technology

Despite all the advances in positive psychology and related fields, and the clear causality between technology and well-being, surprisingly few of these researchers have directly investigated how concrete technologies influence our subjective well-being. The variables emphasized by the researchers typically relate to general human characteristics, such as age, gender, and income—or to events and experiences that do not directly involve technology, such as the effect of engaging in skill-demanding activities (Csikszentmihalyi 1997; Nakamura and Csikszentmihalyi 2009; Seligman 2002), belonging to a community (Demir and Weitekamp 2007; Okun, Stock, Haring, and Witter 1984), perceived meaningfulness (Peterson, Park, and Seligman 2005; Seligman 2002; Wong and Fry 1998), autonomy (Reis, Sheldon, Gable, Roscoe, and Ryan 2000), physical health (Zautra and Hempel 1984), and various forms of sensory pleasure (Kubovy 1999; Reber, Schwarz, and Winkielman 2004; Seligman 2002). Thus, one major challenge lies in translating from empirical findings to concrete technological features.

The first step requires careful attention to the research design and methodology. On the one hand, the empirical findings need to be scrutinized in order to narrow down, as far as possible, precisely what is being measured. The digest version of these findings will often state something like "being social makes you happy," but we need to look into the actual research design and the methods employed in order to see what "being social" actually entails in these cases. For instance, whether the findings are relevant for the value of being social in virtual worlds depends on whether the sociality in question required physical proximity. When the conditions for the positive effect have been identified, the next step is to translate these to concrete technological features—quite often in the form of specific types of interaction. This is probably best illustrated with an example.

"Flow" is a state of mind that has been found to have a profound and lasting boost to well-being (Csikszentmihalyi 1997; Seligman 2002). It is achieved when we are engaged in activities that are challenging and we have the skills to master the challenge. For this to occur, it must be possible to control the difficulty level in order to avoid that it is no longer challenging when we reach a certain skill level while at the same time be manageable when still a beginner. It must also be something to which you can dedicate your whole attention without distraction, and that provides clear and immediate feedback on how well you are doing. Typical examples include the mastery of sports, musical instruments, or hobbies like gardening, dancing, or singing in a choir. If we want to translate this into technological features, the requirements map perfectly onto computer games as an ideal source of flow—when it comes to difficulty settings, immediate feedback, and attentiveness. This is indeed one of the few technologies that have been studied empirically, and the evidence suggests that computer games do have a range of beneficial effects (Baranowski, Buday, Thompson, and Baranowski 2008).

Despite the strong evidence that flow is beneficial for well-being and computer games being ideal sources of flow, this does not mean that computer games are the perfect source of increasing well-being. One problem with translating empirical findings to technological features lies in the fact that the same features may also have negative effects, by themselves or because they necessitate certain other features. The same features that give rise to flow can also give rise to addiction, which is clearly detrimental to well-being. They also necessitate particular practices, such as (in most cases) staying at home in front of a computer screen, which may lead to social isolation—one of the clearest indicators of reduced well-being (Demir and Weitekamp 2007; Okun et al. 1984). The same practice could also lead to deterioration of health, another threat to well-being. Rather than simply concluding that computer games are therefore to be avoided, we can use this constructively instead. For instance, the right question to ask is whether the technology can be modified so that it maintains the flow-capability, yet requires us to simultaneously be social. Gaming technologies like Kinect, Wii, and Rock Band implicitly follow this principle insofar as they require the users to cooperate in physical proximity and this cooperation heightens the sense of flow. This illustrates how the translation between empirical findings and technological features is difficult, and requires us to scrutinize which features are necessary and/or sufficient for the positive effect to occur, whether the same features may be detrimental in some other respect, and whether any such detrimental effects can be remedied by adding, modifying, or removing technological features.

6. Putting it all together: Prudential-Empirical Ethics of Technology

Summing PEET up in one phrase: The purpose of Prudential-Empirical Ethics of Technology is to, retroactively or proactively, evaluate the impact particular technologies have on their users’ and secondary stakeholders’ well-being by carefully translating between empirical research on subjective well-being and concrete technological features, and as restricted by ethical, political, and cultural concerns.

The approach can be summarized as the following step-by-step procedure (although this glosses over many of the nuances involved):

Although this is just a preliminary and general sketch, I believe that a procedure along those lines is necessary in order to arrive at an empirically grounded yet value-sensitive and socially responsible assessment. In the order outlined above, the steps lead to a foundation for developing new technologies. If the purpose is to evaluate existing technologies and their potential impact on well-being, steps 1 and 2 should be reversed and the first step then consists in comparing concrete, technological features with existing empirical results.

The steps, although encompassing a number of issues that require more careful elaboration, serve to illustrate how a robust analysis of the relation between technology and well-being is inherently interdisciplinary. Although psychological research constitutes the empirical grounding, careful attention to ethics and other values is also needed to avoid easy technological fixes to what may be deep-rooted ethical, social, or political problems. The key to approaching such issues is to find an appropriate balance between self-interest (well-being) and other-interest (ethics, politics, and other values), and this is why this project must include both empirical research and philosophy.

7. Advantages, Problems, and Challenges

The advantage with this method lies in the ability to provide concrete and grounded recommendations, thereby avoiding both relativist and paternalist extremes. It is also possible to use this approach with varying levels of refinement, depending on whether it is intended to inspire engineers, inform the public, report to policy makers, or form part of more foundational academic analysis.

As the reader surely has picked up, there are still many challenges and problems facing this approach. First, it rests on a subjectivist notion of well-being that prioritizes first-person experiences. This requires solid justification, and is perhaps the most controversial presupposition from a philosophical standpoint.² Closely related, it also presupposes that empirical research can have philosophical implications, a presupposition that has not gone unchallenged (Feldman 2010).

Second, it can be fiendishly difficult to translate between empirical findings and technological features, in order to safeguard that a non-technological activity retains its beneficial effects when instantiated in a technology. Abstract terms like "being social" and "acts of kindness" are complex, and it will often be challenging to find concrete technological features that give similar positive results.

A third problem concerns the ethical, political, and cultural considerations, where there is clearly a wide range of choices that will determine how often well-being concerns are overridden by other concerns. This is not a problem unique to this approach, however, as most technology assessments presuppose some ethical, political, and/or cultural framework. It can also be seen as an advantage, since it means that the well-being component can be combined with one’s favorite ethical theory without losing the important considerations stemming from the former. Indeed, this approach may be particularly helpful combined with a deontological ethics, since the latter typically gives little guidance when it comes to that which is morally permissible.

Last but not least, it is difficult to find an appropriate balance between applicability and robustness. It aims to be an all-things-considered approach, but the range of things to consider must be balanced against the applicability of the approach—in particular how well it can be appropriated by engineers and policy makers themselves. I plan to develop both a condensed and comprehensive version of PEET, so that the range of necessary considerations is determined by the purpose and target group of the analysis.

8. Concluding Remarks

PEET aims to further the axiological turn by combining the normative, evaluative work that is the hallmark of philosophy and ethics, while using psychological research to empirically ground prudential claims so as to avoid armchair speculation. To avoid any misunderstandings as to the scope of this approach, it should be emphasized that this is not an approach that rivals any other approaches and it is intended to form a complementary form of technology assessment. I am not arguing that we should stop considering issues of right and wrong, but that we should complement such analysis with prudential considerations, and that this should be grounded in empirical research. Allow me to again emphasize that "ethics" is still an integral part of the approach, and that a subjective account of well-being can and (in my view) should be combined with an objective account of ethics, to allow for ethical concerns to override any positive effects on well-being.

I have already presented the approach to numerous engineers, computer science graduates in particular, and I often receive feedback that the approach is, above all, a good source of inspiration for creating novel and creative solutions. This gives me hope that PEET can become an important addition to the ethicist, policy maker, or engineer’s toolbox. The result will be a holistic, ethical, and concrete theory of the role different types of technology ought to have in our lives, of the steps needed to make such an assessment—a framework that presents normative guidelines for engineers, designers, policymakers, parents, caregivers, and others concerned with how technology may or may not contribute to a good life. I have only recently been able to start developing this approach, and I would be very grateful for any criticism and constructive suggestions. At the very least, I hope that this preliminary and cursory overview is sufficient to convey its potential advantage and utility, and to start a fruitful discussion on the role of prudential value and empirical research in ethics of technology.

Acknowledgments

This approach has been in the works for a long time so it is impossible to mention everyone who has helped out in some way. I am particularly indebted to Philip Brey and Pak Hang Wong for important feedback and discussions, along with the participants at the ECAP’11 and WICS’12 conferences. I would also like to thank the many students who have expressed enthusiasm for the approach, along with suggestions for improvement. Thanks, finally, to Peter Boltuc for encouraging me to write up this early outline.

Endnotes
1. Again, there is no room for a full critique in this paper, nor is that my main purpose, but I discuss this in more detail in Søraker 2010.
2. Although a full defence is beyond the scope of this paper, I presuppose a variant of Fred Feldman’s "Intrinsic Attitudinal Hedonism," but one that is "confidence-adjusted" rather than "truth-adjusted." I defend this approach in Søraker 2010, and more systematically in a forthcoming publication.

References
Aristotle. 2009. The Nicomachean Ethics, trans. D. Ross. Oxford: Oxford University Press.
Baranowski, T., Buday, R., Thompson, D.I., and Baranowski, J. 2008. Playing for real: video games and stories for health-related behavior change. American Journal of Preventive Medicine 34(1):74-82. e10.
Brey, P. 2006. Evaluating the social and cultural implications of the internet. SIGCAS Comput. Soc. 36(3):41-48.
Brey, P. 2007. Theorizing the cultural quality of new media. Techné: Research in Philosophy and Technology 11(1).
Brey, P., Briggle, A., and Spence, E.H. Forthcoming. The Good Life in a Technological Age.
Csikszentmihalyi, M. 1991. Flow: The Psychology of Optimal Experience New York: Harper Perennial.
Csikszentmihalyi, M. 1997. Finding Flow: The Psychology of Engagement with Everyday Life. New York: Basic Books.
Demir, M., and Weitekamp, L. 2007. I am so happy ’cause today I found my friend: friendship and personality as predictors of happiness. Journal of Happiness Studies 8(2):181-211.
Feldman, F. 2010. On the philosophical implications of empirical research on happiness. Social Research: An International Quarterly 77(2):625-58.
Fröding, B., and Peterson, M. Forthcoming. Why virtual friendship is no genuine friendship. Ethics and Information Technology (online first).
Griffin, J. 1998. Well-being: Its Meaning, Measurement and Moral Importance. Oxford: Oxford University Press.
Higgs, E., Light, A., and Strong, D. 2000. Technology and the Good Life? Chicago: University of Chicago Press.
Kubovy, M. 1999. On the pleasures of the mind. In Well-being: The Foundations of Hedonic Psychology, edited by D. Kahneman, E. Diener, and N. Schwarz, 134-54. New York: Russell Sage.
Lopez, S.J., and Snyder, C.R., eds. 2003. Positive Psychological Assessment: A Handbook of Models and Measures. Washington, DC: American Psychological Association.
Lyubomirsky, S., Sheldon, K.M., and Schkade, D. 2005. Pursuing happiness: the architecture of sustainable change. Review of General Psychology 9(2):111-31.
Nakamura, J., and Csikszentmihalyi, M. 2009. Flow theory and research. In Oxford Handbook of Positive Psychology, edited by C.R. Snyder and S.J. Lopez, 195-206. Oxford: Oxford University Press.
Okun, M.A., Stock, W.A., Haring, M.J., and Witter, R.A. 1984. The social activity/subjective well-being relation: a quantitative synthesis. Research on Aging 6(1):45-65.
Ong, A. D., and Van Dulmen, M.H.M. 2007. Oxford Handbook of Methods in Positive Psychology. Oxford: Oxford University Press.
Otake, K., Shimai, S., Tanaka-Matsumi, J., Otsui, K., and Fredrickson, B. 2006. Happy people become happier through kindness: a counting kindnesses intervention. Journal of Happiness Studies 7(3):361-75.
Peterson, C. 2006. A Primer in Positive Psychology. Oxford: Oxford University Press.
Peterson, C., Park, N., and Seligman, M.E.P. 2005. Orientations to happiness and life satisfaction: the full life versus the empty life. Journal of Happiness Studies 6(1):25-41.
Reber, R., Schwarz, N., and Winkielman, P. 2004. Processing fluency and aesthetic pleasure: is beauty in the perceiver’s processing experience? Personality and Social Psychology Review 8(4):364-82.
Reis, H.T., Sheldon, K.M., Gable, S.L., Roscoe, J., and Ryan, R.M. 2000. Daily well-being: the role of autonomy, competence, and relatedness. Pers Soc Psychol Bull 26(4):419-35.
Seligman, M.E.P. 2002. Authentic Happiness: Using the New Positive Psychology to Realize Your Potential for Lasting Fulfillment New York: Free Press.
Søraker, J.H. 2010. The Value of Virtual Worlds and Entities – A Philosophical Analysis of Virtual Worlds and Their Potential Impact on Well-Being [dissertation]. Enschede: Ipskamp.
Wong, P.T.P., and Fry, P.S. 1998. The Human Quest for Meaning: A Handbook of Psychological Research and Clinical Applications. Mahwah, NJ: Lawrence Erlbaum.
Zautra, A., and Hempel, A. 1984. Subjective well-being and physical health: a narrative literature review with suggestions for future research. International Journal of Aging & Human Development 19(2):91-110.


Back to top

"Friend" Is a Verb
D.E. Wittkower, Old Dominion University

In the situation described in this passage, surely much of the problem follows from the very short form of the communication possible. Twitter exchanges seem luxurious, indulgent by comparison. But surely much of the problem follows from the format, regardless of length. "Mutual sympathy" is equated here with "flesh and heart," and surely we today agree with Camus that there is a kind of intimacy and connection far easier to establish in face-to-face interaction, or, more accurately, body-to-body interaction (Fortunati 2005, 53), than in writing, no matter whether that writing is limited to ten words. And yet, while the centrality of co-presence and body-to-body interaction might be of unquestionably central concern to erotic relationships, it is far less clear that it should be crucial to friendship. Why, exactly, does writing seem to us to be such a poor substitute for physically co-present interaction within the realm of friendship as well?

The Aristotelian tradition of thought on friendship had accustomed us to dividing friendships between those of virtue, of pleasure, and of utility (and to disparaging the latter two). This puts us on a path to view formal and final causes as determining of friendship—the most prominent distinction between them has to do with their ends, and the friendship of pleasure, concerned so much more with a passing experience than a lasting goal, seems cheap; a guilty pleasure. But friendships with all sorts of formal and final causes have efficient and material causes as well, and these cannot be ignored or discarded.

In terms of formal and final causes we might say that friendship may be based on shared concerns, mutual support in acting on personal and political commitments, principled disagreement and debate, a similar sense of human, shared interests, or common activities and pastimes. In efficient and material aspects, the basis of friendship is found in complaining about the soup, walking together silently, passionate debate, laughing about something unimportant, shopping, playing cards, drinking beer, sharing music, and being bored. Friendship is not a static fact, but a ; part of the active rather than the contemplative aspect of living—and it must be enacted in order to exist.

We are all familiar with the challenges of maintaining a friendship over distances. The practice of friendship, when placed within the context of the longhand letter, must attempt to realize a portion of active life using tools proper to contemplative life. As Vallor puts it,

And not only is shared activity seemingly precluded, but the sharing which is possible tends towards reporting only those aspects of life which appear to us objectively meaningful, removing access to a large portion of what life is to each of us as it is lived, and what would have been shared in a life lived together. "Sharing about" can capture well something of the final cause of friendship—our common values, passions, interests, or humor—but it is hard to see how the written form could contain the material causes of friendship.

A life lived alongside another is disfigured when transferred into narrative: the trivialities of life are no longer a binding connection, but instead the subject of reportage, and the vibrancy of life can only be made again engaging through literary talent on the part of the letter-writer or the empathetic imagination of the reader. As Schopenhauer said, when we view an individual human life objectively in all its many and varied details,

Once we live apart from a friend, and share our lives only through the time-shifted asynchronous written word communicated at a distance, the magnifying lenses of time and space are no longer shared. Our own days are encountered through a microscope, and each moment is one whose passage we feel, no matter how unimportant its content. In speaking to the distant friend, however, we view our life as from afar, and we find ourselves answering simply "Not much" when asked "What have you been up to?" as if we were limited to the space of a telegram as in Camus’s story.

There are certainly friendships that continue and strengthen in important aspects when this transition takes place—Briggle (2008) provides an excellent example of pen-pals who are better able to share meaningful and deeply personal thoughts due to their lack of a shared location. This can be expected to occur in proportion as the friendship has its basis within final and formal causes rather than efficient and material causes; within commitments and projects held in common rather than jokes, movies, and boredom; and, most broadly, within contemplative life rather than active life. But even the friendship based on common values or on intellectual exchange is still a friendship realized in moments lived and shared, and when fully abstracted into an exchange of letters it becomes a placeholder for and shadow of its true form.

Or, at least, this was the case until recently.

I began this section by asking why physically co-present interaction seemed to us to be so central to friendship, and why written interaction seemed to us to be such a poor substitute. Although this fits the way many often think about and talk about friendship, there are surely a great many Facebook users who do not see the dire consequences for friendship that theorists and commentators so frequently bemoan, and many today, markedly but not exclusively tweens and teens, see nothing strange about choosing to IM or text someone who is easily available in person. I don’t believe this is because there’s something wrong with "these kids today," or because we’ve lost the "true meaning of friendship," or because new social media have brought our society into "the shallows." I think it’s because we’re wrong to think of Facebook posts, texts, tweets, and many other forms of new social media communication as written language; or, perhaps less paradoxically put, that writing is increasingly an active and shared activity between distant persons rather than an act of individually composing and subsequently sharing meaning and information.

Changing Communications

Email has been much decried for the way in which the ease, speed, and weightlessness of email communication has resulted in a decline in emphasis upon spelling, grammar, formalities, and etiquette. Moving from email to other new media communications, we see this decline continue, as well as seeing a positive rise in norms of casual, creative, and informal usage, either through convenience (e.g., "ur" for "yours," or the foregoing of apostrophes, which the iOS virtue keyboard puts on a separate screen) or through individual or collective identity construction (as in the case of "l337" typographical conventions or speech patterns stylized after LOLcats, Y U NO guy, or innumerable other Internet memes).

We are not wrong to fear and lament the loss of habits of thoughtful composition and proper language use, even if there are, as I will argue, some positive aspects to this change as well. As the speed and frequency of writing has increased, writing as a practice has become increasingly more functional than thoughtful, and has moved from contemplative to active life. Heidegger even claimed that this process, which he saw at a far earlier stage in the mere act of abbreviating of words (1968, 34-5), threatens the very existence of meditative thought as we use words more and more as tools rather than as bearers of meaning. The phenomenological changes that bring about this danger, through our loss of wider perspectives and our increasing incapacity to step back and consider things carefully and as a whole, bring writing within the realm of the trivial, functional, and immediate. The momentary writing of new social media loses the slow and deep aspects of traditional written thought, but gains the immediacy and vibrancy of momentary speech.

Along with this change in the form and experience of writing, the subjects of communication have undergone a similar alteration. The over-sharing and micro-reporting of the stereotypical constant texter or tweeter exemplify the most mindless and unimportant form of sociality. And yet, it is technologies such as these—and for exactly this reason—that are able, finally, to recreate at a distance those "microscopic lenses of time and space" which allow two people to have the shared experiences that form the material and efficient causes of friendship. This destroys the distance that results in the report, "I didn’t really do anything today," when we have been busy from dawn to dusk with meaningless errands. Through constant tweets and texts, we can be bored and frustrated constantly alongside one another—indeed, the tweeter may very well bring a lived experience of boredom to all his friends, much to their annoyance. As obnoxious as this may be, it does indeed replicate the element of friendship most easily lost in distance.



Here we see a resurgent orality in writing itself—a kind of secondary literacy based in and taking its characteristics from orality; a kind of distorted mirror image of Ong’s secondary orality. Secondary orality is "superficially identical with that of primary orality but in depth utterly contrary, planned and self-conscious where primary orality is unplanned and unselfconscious" (Ong 1977, 298). While secondary orality is oral language derivative of written language, secondary literacy partakes in various and undifferentiated mixtures of literacy and resurgent orality. Some new social media writing is composed with care and concern for meaning, support, and precision in expression, and is altogether in line with our pre-digital expectations for the written word. Some is entirely contextually bound and written extemporaneously from within an ongoing conversation. Most, of course, is some unknowable admixture between the written form and this resurgent orality, and many of the conflicts which take place in new social media can be traced back to different proportions of literacy and orality adopted by those in conversation. One user’s thoughtful and respectful discourse is another’s tl;dr, and both serious communication and sarcastic or absurdist banter fit within the norms of new social media writing.

Facebook Friending

The flexibility and multiplicity of communication structures in Facebook form a platform for the practice of friendship which is able to provide the active experience of proximity seen in texting and tweeting, while providing scalable avenues towards more robust forms of communication and sharing. As Baym makes clear in her discussion of media multiplexity in personal relationships, closer relationships are conducted over more numerous media (2010, 132). Facebook, as a communications platform, can be used in very thin, one-dimensional forms of communication, but its capacity for a variety of different sorts of communication leaves multiplexity as an open possibility within all relationships conducted in part over Facebook.

Status updates supply the opportunity to be present with the other in quotidian trivialities, but do so in a minimally invasive general broadcast rather than a specific communication that might obligate the other to respond. At the same time, the publication of interests, personal details, notes, postings, and videos allows others to hear from one and learn about one’s current interests and concerns, and for one to share with others without having to decide to communicate with anyone in particular. In general, it is a pull-oriented interpersonal communications platform. Rather than the "push" of information outward in the letter, phone call, email, or SMS, a great proportion of the communication that takes place is generated as static content and "pulled down" by viewers of personal pages.

In these ways, the Facebook page is a venue for allowing others to get a sense of the texture of our day-to-day lives, without explicitly inviting anybody to follow it day-by-day. This certainly has its negative aspects, such as the now well-established practice of Facebook stalking, but it also means that there is, in effect, always an open door for our connections to get to know us better.

Targeted communication within one’s network is also always a possibility. Messaging allows for private or personal responses to public conversations, or for conversations requiring intimacy or seclusion—and within this slower, more email-like communicative channel, longer-form writing with a lesser resurgent orality can take place in, as it were, a quiet side-room from the buzz of the friend feed. But more active forms of direct communication are also possible. The "like" button functions much like a head nod or an encouraging "uh-huh" in body-to-body communication. Cuonzo (2010) argues that part of the function of virtual gifts is a kind of post-linguistic social grooming, in keeping with Dunbar’s gossip hypothesis (1997). The purpose, then, of sending someone a flower in a Facebook application, or poking or throwing a sheep at them, is simply to re-establish contact and revitalize a social bond. These things are in some sense meaningless, of course, but so is much of the small talk we engage in regularly, and both serve the same social purpose.

All this, however, approaches Facebook merely as a communications platform. In addition to this, Facebook is able to provide a platform for asynchronous shared activities. To some extent, this is clearly possible through the other communications technologies discussed previously—I may certainly send you a video by email which we can then share our thoughts on, in the same way as I might send you a letter recommending visiting a certain museum—but the multimedia integration of Facebook walls and applications allows for a weightless, immediate, and intuitive approach to distant shared media experiences. Furthermore, the embedded display of linked visual media on the one hand, and the small text-entry windows on the other, both encourage that communications tend to become an asynchronously shared experience of media along with a comment, rather than a written message containing a link to some content referred to. The integration of applications allows collaborative or competitive games and projects to create even more concrete and interactive asynchronous shared experiences.

This is where I disagree with Vallor’s otherwise excellent article, "Flourishing on Facebook." Vallor characterizes our actions in new social media as "communicating to one another information concerning our lives" and "providing the kinds of informational and emotional reciprocity that maintain the will to live together with our friends" (forthcoming, n.p.). I find instead that the writing that takes place in new social media, in line with the idea of resurgent orality, is better characterized as a shared activity rather than mere information sharing. Just as in a conversation in person, what we enjoy is not the information but rather the back-and-forth, the playfulness of banter, and the closeness and connection with our friend, so too do we in new social media often find our conversations to be more about activity than about content. As with in-person conversations, the content itself certainly wouldn’t always (often?) merit the time we spend on it—but that it is time spent with friends does.

Similarly, I see no reason to describe the performance of reciprocity as going no further than a demonstration of the will to live together—is there not enough reason to call it a way of actually living together asynchronously and at a distance, at least within a limited scope of activities? If I post another picture of baby sloths on the wall of a sloth-obsessed friend, I am not indicating something we would do together. The sending of sloths and lolruses is itself a practice within our friendship. If I listen to a lo-fi Mountain Goats video posted by another friend, I do so with the understanding that she listened to it just as I am now, and through my experience of the song I understand more about her experiences and aesthetic sensibility. Every posting can be an invitation to others to encounter the posting as if alongside ourselves, and in this way we can have a meaningfully robust asynchronous version at a distance of many of the everyday material causes of friendship: watching TV or a movie, discussing the news or an article, shopping together, playing Scrabble, and so forth. Through the limited forms of cellphone and camera photos and videos, we can even, in a limited and thin way, invite our network to visit our family, or keep up to date on our vacation (or a party) as it happens. Now, Vallor is concerned specifically with complete friendships and friendships of virtue, so she has other reasons to dismiss these activities as relatively trivial to her primary concern, but I think it important nonetheless to recognize that new social media provide us not just with a means of communication but a means of living together, and, as Vallor well recognizes, friendships of virtue usually come about as a deepening of relationships initially established in the pursuit of utility or pleasure.

Old Friends, New Media

To see how new social media are relevant to the possibility of deepening relationships towards virtue, consider the difference between making contact with old friends via Facebook versus via email, telephone, or letter.

In friending someone, there is a very minimal social commitment. Users often seem to use friending as a means of recreating former social groups, and will friend those with whom they have no immediate desire to communicate, but just want to check in on, touch base with, or keep in touch with. This surely has some negative aspects—users tend to accumulate friends who they know and care very little about—but the lowered expectation allows people to establish connections with those they are very much out of touch with.

The case of the old friend to whom we have little to say exemplifies the most characteristic strengths of Facebook as a communications platform and as a platform for asynchronous shared activity. In this kind of relationship, contact would be unlikely in other media, for we would not have sufficient initial desire and purpose to motivate the investment of time and feeling requisite to reestablish a meaningful relationship. The low commitment of friending allows this contact to be established, which then, through status updates and news feeds, gives us a sense of who our old friend has become, and of what the course of her life now consists. We may find we share interests, commitments, and projects—either those that once bound us together, or others that have arisen in the intervening period. We may find we have more in common than expected, or we may rediscover, for example, that shared sense of humor which had always been the only thing we had in common.

And, on Facebook, a shared sense of humor and fun is enough to re-establish a relationship—and, even more remarkably, a relationship that can grow. Simple activities with scalable levels of interaction, such as Facebook games, allow us to move smoothly from playing a game because the game is itself diverting to playing a game as a way of being together with a friend. Competition and chatting as we play allows a return to the quotidian moments of passing time which once brought us together, without the artificial attempt to reestablish a friendship using means foreign to the shared activities which formed an initial, now lapsed bond.

In these ways, Facebook is a remarkably well-suited platform for the activity of friendship. Where the connections forged are superficial, they allow avenues for growth and intensification. Where the messages and posts are terse and simple, they allow for conversations and shared experiences to emerge. Where the group associations and activities are thin and basic, they allow opportunities to raise awareness and recruit others to causes that may become passionate commitments.

To be sure, the movement of communications technologies from the realm of contemplative to that of active life presents cultural problems and dangers, but it allows the long-distance elements of friendship to become not a mere sharing of information about activities engaged in separately, but an active asynchronous sharing of activities themselves. This active component of new social media communications is both important and easily misunderstood, and as long as we do not grasp that communications here are not reportage and summary, but asynchronously shared experience at a distance, much of what happens in new social media will appear to be useless self-important triviality. We can call this the "sandwich problem."

Why do people talk about their sandwich? Why do people post pictures of lunch? Surely it is not because they are under any illusion that their sandwich is of significant objective importance. Neither, if we are to be charitable, can we assume that it is because the sandwich-sharer believes that he is of such great importance to friends and associates that the slightest and most uninteresting details of his life take on interest by association, like Catholic relics touched by saints. What, then, can be the motivation?

This will be ever a mystery to us as long as we believe that the point of the communication is the information it contains. The point is to invite friends to lunch. The sharer can then eat alongside his absent friends, who he knows to be experiencing the appearance of his sandwich, as if sitting across the table. The friends, for their part, have been granted this window into the life of their friend, which they may either ignore entirely, or choose to reflect upon and, in so doing, revitalize their connection. If this kind of asynchronous, opt-in broadcast, mediated togetherness-at-a-distance displaced more robust forms of shared activity, this would certainly be a worry—but the point is that we can’t even make sense of why people do this unless we accept that this is experienced as a shared activity of communion, rather than a communication of information.

"Friend" is a verb, now, thanks to Facebook—and it is well that this is so. While "friending" is the mere establishment of a connection on this social networking site, what is both most important and most surprising about Facebook is its affordances for friendship as an activity.

References
Baym, N. 2010. Personal Connections in the Digital Age. Malden, MA: Polity.
Briggle, A. 2008. Real friends: how the Internet can foster friendship. Ethics and Information Technology 10(1):71-79.
Camus, A. 1991. The Plague, trans. S. Gilbert. New York: Vintage.
Cuonzo, M. 2010. Gossip and the evolution of Facebook. In Facebook and Philosophy, edited by D.E. Wittkower. Chicago: Open Court.
Dunbar, R. 1997. Grooming, Gossip and the Evolution of Language. Cambridge: Harvard University Press.
Fortunati, L. 2005. Is body-to-body communication still the prototype? The Information Society 21:53-61.
Heidegger, M. 1968. What is Called Thinking? New York: Harper & Row.
Ong, W. 1977. Interfaces of the Word. New York: Cornell University Press.
Schopenhauer, A. 2007. The vanity of existence. In Parerga and Paralipomena, trans. T.B. Saunders. New York: Cosimo.
Vallor, S. Forthcoming, n.p. Flourishing on Facebook: virtue friendship & new social media. In Ethics and Information Technology. Published online: http://www.springerlink.com/content/k7754743398g3k65/ (Jan. 8, 2011).


Back to top

Embodied Cognition and the Turing Test: An Uncomfortable (Re-)Union
Robin L. Zebrowski, Beloit College

Abstract

For decades, the artificial intelligence community has questioned the validity and strength of the Turing Test as a way to evaluate the presence or absence of a mind. Given the recent surge of evidence for embodiment theories, the Turing Test has been considered largely irrelevant. However, conceptual metaphor theory, a strong theory of embodiment, ironically offers a way to save the Turing Test. In spite of the fact that the theory, as articulated by Lakoff and Johnson (1987, 1999) is explicit in its rejection of AI, it offers a strong linguistic basis for evaluating the presence of an underlying mind.

Keywords

Embodied cognition, Turing Test, embodiment, conceptual metaphor, cognitive semantics, artificial intelligence

It is almost impossible to over-estimate the impact that Alan Turing’s 1950 paper Computing Machinery and Intelligence has had on the artificial intelligence (AI) community. Turing’s historic paper set the stage for a sort of "put up or shut up" moment in the field, challenging those who spend their time working out what it might mean for a machine to exhibit genuine intelligence, and setting the bar for AI theorists indefinitely. And while Turing almost certainly never intended for his Imitation Game (what has come to be known as the Turing Test) to be an operational definition (that only and exactly those things that pass it count as intelligent), the test took on a life of its own over the sixty-plus years since Turing proposed it. In fact, I know of almost no one in the field of AI theory who defends the test as written (excepting maybe Daniel Dennett, whose defense will be discussed below), and yet it remains the benchmark that everyone almost-embarrassingly references, seemingly dismissing it as overly simplistic while using the same breath to grudgingly admit that no machine or program has yet passed it. It is this apparent contradiction that I intend to discuss here, while offering evidence and argument that the Turing Test is overdue to be re-examined as a potentially valid test for artificial intelligence.

Despite the Turing Test being generally the most talked-about and well-known test for AI, most discussion in the literature has remained critical. Particularly in light of what I will call the Embodied Revolution in the fields of AI and cognitive science, the Turing Test has been pushed even further from prominence in recent years. As more and more researchers embrace the claims of embodiment theories, a purely linguistic test begins to look less and less viable as a candidate for a true test of intelligence. By "intelligence," I include the concept of mind or consciousness that is the true goal of AI. (It isn’t clear to me what intelligence looks like in the absence of a mind, since arguably we have many machines that are already intelligent in that sense and yet we do not believe we have achieved true AI. Searle’s (1980) distinction here between Strong and Weak AI is relevant, and I’m interested only in the notion of Strong AI here.) In what follows, I will give the reader a brief background of AI’s history, including the role played by the Turing Test, spending some time on the Embodied Revolution and what it has meant to the field. Then, I will argue that Lakoff and Johnson’s work on Conceptual Metaphor specifically (and ironically) offers us a reason to re-think this move away from the Turing Test. Ultimately, it seems as though one can reject many of the assumptions of what Haugeland (1986) called "GOFAI" (Good Old-Fashioned AI), including that human thought is primarily symbol manipulation, and still accept that the Turing Test is a valid test of human-like intelligence.

Historically, AI has gone through a number of research paradigms. Different philosophers have carved this history up differently, but one can see fairly clearly that there have been at least three major approaches to the field, all of which overlap one another chronologically at some point, and all of which remain active research projects today. We often use the term "GOFAI" to describe the earliest work in AI, that which is based on the Physical Symbol System Hypothesis of Newell and Simon (1976). In their words:

It’s not difficult to see how the Physical Symbol System Hypothesis approaches the problem of intelligence in AI much the same way Turing did, in so far as the focus remains firmly on the symbolic representation of information. Although Turing wanted to limit the test to only digital computers, the Turing Test was less concerned with the specific underlying mechanism of intelligence, presumably beyond symbol manipulation itself, as Turing was intentionally moving away from trying to define what intelligence actually is. However, language is itself a symbol system, and it is the symbol system with the deepest relationship to human thought. GOFAI and the Turing Test were very closely tied, to the point that around the 1960s, one could see both programmers in AI as well as philosophers and theorists all working specifically on how to get a machine to pass the Turing Test, while discounting most other questions about the nature and structure of intelligence itself (as Turing intended).

The second of the three major research paradigms in AI is connectionism, and all of the various forms it took. While the history of the similarities and differences between connectionism and GOFAI is fascinating (and we might even say there was a connectionist revolution at one point) it isn’t entirely relevant to my discussion here today. It is worth noting, however, that when the connectionist approach did challenge the GOFAI approach, the primary arguments about what connectionism could and could not do were primarily focused on the failures of connectionism with regard to language itself (Fodor and Pylyshyn 1988; Fodor and McLaughlin 1990; McLaughlin 1993; Smolensky 1988, 1994). It seemed, for a very long time, that no connectionist system would be able to process natural language at all, let alone ever pass the Turing Test.

Then, in the late 1980s, Rodney Brooks challenged both of these traditions and made an argument that, in principle, the Physical Symbol System Hypothesis was incorrect. Instead, Brooks claimed that the body itself had primacy in intelligence, and that in order to understand and replicate human intelligence, we needed to start with simpler intelligences that lacked symbolic representation in the traditional fashion altogether (1991). He argued that "mobility, acute vision and the ability to carry out survival related tasks in a dynamic environment provide a necessary basis for the development of true intelligence" (141). In a 1999 anthology of his work, Brooks reports that this now-historic paper ("Intelligence Without Representation") was turned down by journals for four years before finally being published in the most well-regarded journal in the field. This historic fact is important because it shows that there was widespread resistance to this move toward embodiment, and it is here that I would argue the Embodied Revolution in AI truly began. It is worth noting that one can—and many do—trace the primary arguments in Brooks’ work back through many earlier philosophical movements, and many of his claims can be found in the work of the phenomenologists like Merleau-Ponty and Heidegger, and even in some of the early American Pragmatists, like Dewey (Gallagher 2009). I place the beginnings of the Embodied Revolution here only as it relates to Artificial Intelligence. It arose almost independently across various fields of cognitive science at other points in history.

When I refer to "embodiment," or refer to the mind as "embodied," I mean simply that the structure of our bodies is a primary determinant of the possibilities of our cognition. According to Tim Rohrer, "the embodiment hypothesis argues that minds are fundamentally not disembodied algorithmic processes like those in a (serially-driven discrete state) digital computer program, but are instead constituted and constrained by the kinds of organization reflected in the biological, anatomical, biochemical, and neurophysiological characteristics of the body and the brain" (Rohrer 2001a, 3). While the roots of contemporary Embodiment Theories reach easily back through phenomenology and pragmatism, we might argue that the philosophical Embodied Revolution began in earnest with the publication of Lakoff and Johnson’s Metaphors We Live By in 1980. (These dates and originating works will almost certainly be debatable amongst researchers depending upon one’s goals, and my back-of-the-envelope calculations of where these movements began can be put aside as irrelevant to my greater point.) Metaphors We Live By was the text that originated Conceptual Metaphor Theory, which, in its simplest form, claims that much of our conceptualization is structured metaphorically, from aspects of our physical, cultural, and social environments. What this means is that the very possibility of my acquiring some concepts rests first on the way my body is structured and how it interacts with my environments. In other words, metaphor is not merely a matter of language, but a matter of thought itself. To take one of the most famous examples, we use our sensory experiences with vision to make sense of the abstract concept of knowledge itself. (This is the KNOWING is SEEING metaphor). We can see examples of this in our language about knowledge, such as in the following common phrases: "I see what you’re saying," "I see what you mean," "It looks different from my point of view," "The argument is clear," "It was a murky discussion," "I view that differently" (Lakoff and Johnson 1999, 54; Lakoff and Johnson 1980, 48). This is one of many hundreds of conceptual metaphors that have been dis-covered (a word which itself is almost certainly related to this metaphor).

The field of embodied cognition, though slow to get a foothold, has grown across all relevant domains in recent years. It is fairly safe to say that almost all fields related to mind or cognition now have sub-fields working on embodiment theories. The Embodied Revolution has been robust, and work in the philosophy of mind, neuroscience, linguistics, artificial intelligence, robotics, biology, and psychology (at least) now exists to both explore and support the embodied mind hypothesis. As a challenge to the last 2,000 years of metaphysical foundations in science and philosophy, embodied mind theories can and are changing the way we think about everything from ourselves to the world itself.

Importantly, Conceptual Metaphor Theory offers a challenge to the functionalist theory of mind that underpins the entire field of artificial intelligence. The theory shows us that we often reify abstractions in order to talk about them and make sense of things that otherwise elude our senses. Lakoff and Johnson sometimes call these "ontological metaphors," and argue that viewing an abstraction as an entity allows us to "refer to it, quantify it, identify a particular aspect of it, see it as a cause, act with respect to it, and perhaps even believe that we understand it" (Lakoff and Johnson 1980, 26). Importantly, of course, since these are ways that we conceptualize things, we rarely recognize any of this as metaphorical: we understand the mind, for example, as a thing precisely for these reasons, and it is precisely because this is useful that we fail to realize that this is not necessarily any sort of objective truth about the mind. (Indeed, we can see the different sorts of research projects that follow from fundamentally different ontologies of mind; we might have decided mind was a process rather than a thing, which brings with it very different assumptions and research methodologies.) Lakoff and Johnson have argued that the field of AI is largely a result of at least three metaphors that Anglo-American analytic philosophy has given us. First, the THOUGHT as LANGUAGE metaphor. This will be familiar to anyone trained in traditional analytic philosophy because of Chomsky and Fodor’s arguments that claimed thought literally was a kind of language (Fodor 1975). Some of the language we have that reveals this underlying conceptualization can be found in the following common phrases: "Let me make a mental note of that," "I can read her mind," "He’s reading between the lines," and "Her thoughts are eloquent" (Lakoff and Johnson 1999, 244-245). We can see Newell and Simon’s Physical Symbol System Hypothesis as a clear result of the belief that thought has the properties of language, although of course they are making a literal claim where Lakoff and Johnson are arguing it is metaphorical conceptualization.

We might look to another metaphor for mind: THOUGHT is MATHETHAMTICAL CALCULATOR. With this conceptualization, we understand reasoning to be like addition, for example. We might say of thought, "I put two and two together," or "It doesn’t add up" (Lakoff and Johnson 1999, 246). Again, both Newell and Simon’s physical symbol systems and Turing’s own understanding of human-level intelligence as symbol manipulation can be seen in this metaphor. If numbers can be represented by sequences of written symbols, by this understanding of a mind, so can thoughts (247). The Turing Machine itself comes into play here.

Finally, of course, this leads to the most important conceptualization for our purposes: the MIND as MACHINE metaphor. It must be noted that within the field of AI, most of us do not believe this conceptualization of a mind to be metaphorical at all. It is in some ways obvious that this is taken to be literal, and some researchers are explicit about this (Pinker, 259). The MIND as MACHINE metaphor plays out in the following way: the physical computer is used to understand the person (and especially the brain); we understand the mind to be like a computer program, concepts to be formal symbols, the conceptual system to be like a computer language, thoughts to be like formal symbol sequences, thinking to be formal symbol manipulation, memory to be a database, knowledge to be the contents of that database, and the ability to understand is the ability to successfully compute (257). Lakoff and Johnson see this mapping of mind onto machine as merely metaphor, and hence do not see AI as a viable research project. While I agree that this conceptualization is metaphorical and that our minds may not be machines in this literal sense, I do not see this as the death of the AI project. In fact, the work in embodied AI of people like Brooks and Pfeifer shows us that we can accept this conceptualization as metaphorical and still continue working on artificial intelligence with no apparent contradiction (Pfeifer and Iida 2004).

Here is the great irony of Lakoff and Johnson’s work in conceptual metaphor theory: while they argue that it shows the error of AI in general (and the misguidedness of physical symbol systems like those on which Turing based his famous Imitation Game), I argue that they have offered us a valuable reason to re-consider the Turing Test as a valid test for artificial intelligence. It is worth pointing out that there is a tremendous amount of independent evidence for conceptual metaphor theory at this point, across an incredible number of different fields and domains of knowledge (Lakoff and Johnson 1999). In spite of the fact that we can now see evidence of these claims in neuroscience and experimental psychology (for example), and using a number of different methodologies, it is, in an important way, still a linguistic theory. The structure of our deeper levels of conceptualization is revealed through our language usage, even though that language usage isn’t itself the structure of thought. When I ask someone whether they truly grasp my idea, they must necessarily understand the word "grasp" metaphorically, in an embodied way, from how they would physically grasp an object. In fact, there is neural evidence that our minds arose evolutionarily by piggy-backing on the sensory-motor neurons already in place for sensing and movement, using the same structures to understand abstract ideas rather than building new neural structures from scratch (Lakoff and Johnson 1999). So, when I ask someone if they "grasp" what I’m saying, the evidence shows that our brains are activating the neural area for our hands themselves, which first and foremost do our grasping for us (Rohrer 2001b). Therefore, one necessarily must have grounded that physical experience first before being able to use the term in a metaphorical way. (There’s something to be said here for a solution to the symbol grounding problem, as well, which I will return to shortly.) There is evidence from developmental psychology that we do, in fact, go through these stages of usage developmentally, including a stage of conflating the literal and the metaphorical usages before properly using the metaphorical form correctly (Lakoff 2008). If this theory of conceptual metaphor is true, and I think we have, in the last three decades, acquired enough evidence to believe that some version of it is, then our language really does reveal something deep and important about the underlying structure of our minds. In fact, without natural languages, we would almost certainly have no access to this information about how minds work.

And now it’s time to bring this back to the very beginning: the Turing Test. It is true that the Turing Test has been dismissed widely in the last few decades as not being a proper test of intelligence (including arguments that we will be able to "brute force" a win before we have reason to believe there is genuine intelligence, and arguments from the other side that intelligence will arise way before the test is passed). One interesting revision of the Turing Test comes from Steven Harnad, who formulated the symbol grounding problem, illustrating that symbols need to be given content in a way that is meaningful for the system itself. Over the last several decades, Harnad has offered revisions of the Turing Test in order to differentiate various levels at which a system might be capable of acquiring meaningful symbols. He has laid out a hierarchy that stretches from the toy level Turing Test (T1) that of the Loebner competition, for example, to the test as Turing formulated it (T2), to a version where the system has human-level sensori-motor capacities via robot (T3), and even to a neural level model (T4) (Harnad 1989, 1995). Harnad ultimately seems to argue that T3 (robotic capacities) is the appropriate level at which we need to aim if we want to be able to assume the test is being passed in the appropriate way (with grounded symbols). However, it seems Harnad’s argument differs from mine in an important way: he is claiming that intelligence is more than a linguistic capacity, and as such the robotic level of the Turing Test is the appropriate one because humans do more than linguistic things. While I agree wholeheartedly with Harnad, our goals are somewhat different. He is illustrating the problem with the traditional version of the test itself, and giving us a more rigorous version of the test that involves other sorts of intelligence that humans and other animals certainly have. I agree with Harnad’s claims with regard to this revision. However, my own project is to show that if we want to restrict ourselves only to Turing’s formulation of his test, Harnad’s T2, then passing the test as Turing articulated it is still a powerful proof. I just happen to think it cannot be done without there also being a fully embodied creature with grounded symbols. Harnad’s argument and my own converge in practice, but are distinct on the level of theory.

One notable exception to the overwhelming number of critiques of the Turing Test has been Daniel Dennett’s 1984 paper, in which he defended it on the basis of world knowledge. In this piece, Dennett claims that the Turing Test is a strong test of an underlying mind, but not because mind is, at heart, nothing but pure symbol manipulation. Rather, Dennett argues that the test will only be passed by a creature who has a tremendous amount of background information about the world and can parse ambiguous language in complex contexts. In other words, the test works because language and the world are both highly complex systems, and the Turing Test taps into that complexity in a tightly controlled way. For example, Dennett offers the distinction between the following two sentences, originally identified by Terry Winograd, that we who live in human societies can parse easily enough. Note there is only a single word that differs between them, but that the meaning is entirely changed: 1) "The committee denied the group a parade permit because they advocated violence" and 2) "The committee denied the group a parade permit because they feared violence" (Dennett 51). As Dennett points out, the difference in the verbs (advocated and feared) changes who the pronoun refers to. Both are legal parsings, but no one who understands what a parade permit is would likely misinterpret the pronoun. Dennett’s world knowledge argument is an interesting one, and it’s not clear to me that there isn’t some accidental overlap with my own argument. However, there is no clear understanding of what Dennett’s "world knowledge" is, how one can measure it, or whether it isn’t simply another thing that can be "brute forced" in a sufficiently complex system. (We might consider looking to IBM’s Watson for a candidate system with apparent world knowledge and nothing we would consider genuine thought. It is worth pointing out that Watson is a long way from passing the Turing Test.) Dennett accounts for this somewhat by pointing out that he believes the test is powerful enough that if bodies are genuinely needed for minds, then nothing will pass the Turing Test without a body, but he leaves it an open question. He also claims it is an empirical question. I’m arguing that we have empirical evidence that comes down clearly on the side of embodiment, and that it is conceptualization itself rather than world knowledge that matters.

Therefore, I believe Conceptual Metaphor theory gives us a precise methodology¹—a body sufficient like ours—and plenty of specific examples that would sufficiently indicate that a system or person had some sort of genuine understanding of even very abstract concepts. (See, for example, the conceptual metaphors for such things as LOVE is a JOURNEY; IDEAS are FOOD; KNOWING is SEEING, to name a few.) Ironically, embodiment theory alone would seem to preclude the Turing Test from being a valid test of intelligence, given that it is a purely linguistic test, except that Conceptual Metaphor Theory has shown us that language is our very window into that mind.

In the process of shifting our best practices in AI research to embrace embodiment theories, many of us have tended to throw out the Turing Test as a remnant of an outdated view of minds. However, we’ve thrown the metaphorical baby out with the metaphorical bathwater; language itself is still our best tool with which to uncover hidden minds—a shortcut of sorts. And while this still cannot account for the possibility of minds without language (in which I of course see no inherent contradiction and plenty of real world examples), it gives us a strong way to discover if there’s a genuine mind in any sort of creature that has language.

Endnotes
1. It is probably worth an aside to mention that I cannot say here what "sufficiently like ours" means. It is clear that our social, physical, and cultural environments all seem to play a role in our embodied existence, and things we might take for granted like gravity, or the fact of our having fronts and backs and tops and bottoms, are all relevant. I do not want to imply that there is a single sort of human body (although I think Lakoff and Johnson are claiming this). There is real human variation and it matters for us, but the degree to which we must have bodily overlap, and what aspects must overlap, is not yet apparent. I think this is an empirical question, but a lack of critical examination of the concept of embodiment itself has kept us from even recognizing the importance of the question. Wittgenstein may have been exactly right when he wrote of talking lions and our inability to communicate with them.

Bibliography
Brooks, R. Cambrian Intelligence: The Early History of the New AI. Cambridge, MA: MIT Press, 1999.
Dennett, D. "Can Machines Think." In How We Know, edited by M. Shafto. San Francisco: Harper & Row.
Fodor, J.A. The Language of Thought. Cambridge, MA: Harvard University Press.
Fodor, J.A. and Pylyshyn, Z. "Connectionism and Cognitive Architecture: A Critical Analysis." In Connections and Symbols, edited by S. Pinker and J. Mehler. Cambridge, MA: MIT Press. (A Cognition Special Issue).
Fodor, J.A. and McLaughlin, B. "Connectionism and the Problem of Systematicity: Why Smolensky’s Solution Doesn’t Work." Cognition 35 (1990): 183-204.
Gallagher S. "Philosophical Antecedents of Situated Cognition." In The Cambridge Handbook of Situated Cognition, edited by P. Robbins and M. Aydede. 35-53. Cambridge, MA: Cambridge University Press.
Harnad, S. "Minds, Machines and Searle." Journal of Theoretical and Experimental Artificial Intelligence 1 (1989): 5-25.
Harnad, S. "Grounding Symbolic Capacity in Robotic Capacity." In Steels, L. and R. Brooks (Eds.) The "artificial life" route to "artificial intelligence." Building Situated Embodied Agents, edited by L. Steels and R. Brooks. 276-86. New Haven: Lawrence Erlbaum, 1995.
Haugeland, J. Artificial Intelligence: The Very Idea. Cambridge, MA: MIT Press, 1985.
Lakoff, G. and Johnson, M. Metaphors We Live By. Chicago, IL: University of Chicago Press, 1980/2003.
Lakoff, G. and Johnson, M. Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought. New York: Basic Books, 1999.
Lakoff, G. "The Neural Theory of Metaphor." In The Metaphor Handbook, edited by R. Gibbs. 17-38. Cambridge: Cambridge University Press, 2008.
McLaughlin, B.P. "The Connectionism/Classicism Battle to Win Souls." Philosophical Studies 71 (1993): 163-90.
Newell, A. and Simon, H. "Computer Science as Empirical Inquiry: Symbols and Search." Communications of the ACM 19, no. 3 (1976): 113-26.
Pfeifer, R. and Iida, F. "Embodied Artificial Intelligence: Trends and Challenges." In Embodied Artificial Intelligence: International Seminar, Dagstuhl Castle, Germany, July 7-11, 2003, Revised Selected Papers, edited by F. Iida, R. Pfeifer, L. Steels, and Y. Kuniyoshi. Heidelberg: Springer, 2004.
Pinker, S. The Stuff of Thought: Language as a Window into Human Nature. New York: Penguin Press, 2007.
Rohrer, T. "The Cognitive Science of Metaphor from Philosophy to Neuroscience." Theoria et Historia Scientarium 6, no.1 (2001a): 27-42.
Rohrer, T. "Understanding through the Body: fMRI and ERP Investigations into the Neurophysiology of Cognitive Semantics." Paper presented at ICLC 2001. UCSB, Santa Barbara, July 2001b.
Searle, J. "Minds, Brains and Programs." Behavioral and Brain Sciences 3, no. 3 (1980): 417-57.
Smolensky, P. "On the Proper Treatment of Connectionism." Behavioral and Brain Sciences 11 (1988): 1-23.
Smolensky, P. "Connectionism, Constituency, and the Language of Thought." In Meaning in Mind: Fodor and His Critics, edited by B. Loewer and G. Rey. Oxford, UK: Basil Blackwell, 1990.
Smolensky, P. "Constituent Structure and Explanation in an Integrated Connectionist/Symbolic Cognitive Architecture." In The Philosophy of Psychology: Debates on Psychological Explanation, edited by C. Macdonald and G. Macdonald. Oxford, UK: Basil Blackwell, 1994.
Turing, A.M. "Computing Machinery and Intelligence." Mind 59 (1950): 433-60.


Back to top

Selves in Video Games: Reflections by Two Middle-Schoolers in an After-School Program on Consciousness
Cole Sohn and Maxwell Mainman, Saint Gregory Preparatory Middle School, Tucson, Arizona
Iris Oved, University of Arizona

1. Introduction

Take a moment to remember playing some of your favorite video games. Maybe you loved Pac Man as a teenager, or Super Mario Brothers, and maybe now you play World of Warcraft or Second Life. If they are good games, perhaps you would say that you are able to escape into them, maybe for hours at a time. Did you ever stop to reflect on this phenomenon of "escaping into" a video game? The first two authors of this paper did, at age 12, when asked to come up with questions to explore during an after-school program called Puzzles About Consciousness. The questions they raised were: (1) Where are you when you are playing a video game? (2) Are you a different self when playing different video games? (3) Does how engaging a video game is influence where you are and who you are when playing the game? The next few sections of this paper were written by the kids, in exploration of these questions.

First, a few words about the after-school program. This program was developed by the Paradox Center, which is in the process of becoming a non-profit organization with the aim of teaching and studying creative critical thinking in the youth. We met once a week, from January to May of 2012, at Saint Gregory Preparatory Middle School in Tucson, Arizona. For the majority of the sessions, we used methods designed by the Philosophy for Children movement, in which students are given a stimulus (a movie clip, a picture, a piece of a story, to name a few), and the facilitator either raises a question for the students to explore or allows the students to come up with questions and vote on one or two for group discussion. The students then take turns contributing to the inquiry, calling on one another after making their moves, and labeling their moves in the inquiry (hypothesis, reason, example, distinction, analogy, clarification request). The facilitator is there to ensure that the inquiry is fair and on topic and otherwise contributes only minimally to the content. The aim is to give students the opportunity to follow their own path of inquiry. For the majority of the sessions in this after-school program, we discussed well-known thought experiments from the consciousness literature, simplified, kiddified, and shortened. Topics included color inversion (originally from John Locke, 1689), Daniel Dennett’s "Where Am I?" (1978), Frank Jackson’s Monochromatic Mary (1986), and a video of a robot with human-like skin, coupled with the question, "If you discovered that your granny was a robot, would you still think she loves you?"

In the final few weeks of the after-school program, two of the students (the first two authors of this paper) started developing ideas for a poster presentation in the international Toward a Science of Consciousness conference, which was being held in Tucson (as it is biennially) in April of 2012. For the poster, the students were asked to come up with their own questions, hypotheses, reasons, and "why I might be wrong." This paper develops the ideas presented in the poster. They mostly focused their discussions on the video game Minecraft.

2. Description of Minecraft (written by Maxwell Mainman and Cole Sohn)

Minecraft is a computer game in which everything is made of blocks. Everything is pixelated, and cubed. It is a randomly generated, open world game where there is no story line. You may play in single player by yourself, or multiplayer with others. Separate programmers create expansion packs, called mods, to the original game. The Minecraft world is designed like the real world with minor exceptions. You start out by cutting down trees to get wood to make tools and items. As you move along you upgrade your tools, weapons, and items to be better at their tasks. You use these tools and items to complete the goals of the game. The game has many goals, but they all lead up to building great things. The side goals include mining, crafting, building, farming, fighting, and exploring.

Mining consists of going through cave systems and abandoned mines while using your pickaxe and mining out ores. Crafting is when you combine lesser items to create greater items on a special bench called a "crafting bench." Building is when you place material blocks to create structures. Farming in Minecraft is like farming in real life. You prepare the ground with tools, plant seeds, wait for them to grow, and harvest them when they’re ready. There are randomly generated monsters roaming your world. You can destroy the monsters with various weapons and get items that are only attainable by killing them. You can explore your world and find different biomes with different resources, advantages, and disadvantages.

3. Where Are You When Playing a Video Game? (written by Cole Sohn)

I thought about this question after thinking about "Where is Daniel?" A guy named Daniel has to leave his brain in a container, connected by radio waves to his body, and then travel underground with his body. I started to wonder, where do you exist when you are playing a video game like Minecraft?

I think it depends on whether your mind determines where you are or your body does. My hypothesis is that when playing video games, your mind is partly in the video game world and partly in the real world, and more of you is in the game if you are more engaged by it.

In support of the idea that part of your mind is in the video game world, I have observed that sometimes when you are playing some video games, your bodily actions, sensations, and emotions coincide with what your character would feel if he were real and in the game world. For example, in Minecraft, when you find a special item like diamonds and get back to your home you feel happy and relieved; when you lose your valuable items in Minecraft, you feel upset and down; also, when you are falling off a cliff, you feel a slight vertigo. The more engaged you are by the game, the more extreme the experience is.

Despite all of this, I think that part of your mind remains in the real world. You can still talk to people in the real world and you know what is going on around you. It is like dreams. You don’t always know that you are dreaming. Sometimes you think that it is real. This means there is a higher percentage of you in the dream than there would be of you in the video game. Also, the more options the game has, the more engaged you are, and the more you are in the game.

I might be wrong. People might think your mind is always in the same place, where your body is. But I believe that the world you are experiencing is where your mind is, so if you are experiencing a video game world, your mind is in the game world, and when you are experiencing the real world, your mind exists in the real world. When playing a video game, you are experiencing both the real world and the video game world, so your mind is in both. I don’t think that when you experience a table your mind is in the table. It is the world you are experiencing that determines where your mind is.

Another way I might be wrong is if people think you are completely in the video game or dream because that is where you think you are at the time. I think that you are not completely in the video game or dream because you know it’s a video game and at least sometimes you know it’s a dream. Also, things from the real world affect you when you’re playing a video game, and also they can affect your dream. For example, if you hear a loud noise, something could happen in your dream that has a similar noise. More of you exists in a dream than in a video game because less of the real world affects your experience.

4. Who Are You When Playing a Video Game? (written by Maxwell Mainman and Cole Sohn)

Who are you when playing a video game, and are you a different self while playing different video games? Our hypothesis is that you are a different self in different games because of the environments and consequences but you retain core aspects of your real-world self like morals and values. The "game you" sometimes reflects your personality but it is more exaggerated. It also sometimes reflects what you wish you were able to do in the real world.

One thing we have observed is that you are a slightly different self in a video game because video games have very different environments from the real world and from other games. Because the environments are different, there are different things that you can do, and different things that you can experience.

Even though some aspects of your personality change when you play a video game, core aspects stay the same. One observation is that when some people choose the character at the beginning of a game, they choose the same features in different games. For example, one of us (Max) is a dwarf in Minecraft and Dungeons and Dragons, and prefers to use dwarven weapons in Skyrim. Also, skills remain constant across games (such as skill with a bow and arrow). Dwarfs usually have similar abilities in each game, and that allows the player to keep the same traits. It is a bit different with dreams. You have different abilities in a dream, like flying, but you are still yourself in your dreams.

There are some reasons to think we are wrong about this. People who cheat in games may not cheat in real life, like on tests. Also, people could be mean and aggressive in a game but in real life they are kind and shy. This seems to suggest that some core values may change when playing a video game. But, maybe this is just because they are worried about consequences in real life. Being aggressive would still be their core value. Another observation is that when playing video games you are more adventurous than you are in real life. But maybe this is just because in real life you have real risks.

5. Options, Engagement, and Escaping Into the Game (written by Cole Sohn and Maxwell Mainman)

We think that the more engaged you are in a video game, the more you exist in the game world and the more you become the character. We think that the more options you have in the game the more engaged you are (options about what the character does, not about the consequences). A major example of this is role-playing games (RPGs). You become more creative, you have more to think about and consequences to consider.

To explore this hypothesis, we considered many different kinds of games, and surveyed our friends, and often the more engaging ones are the ones with more options. In Open-world games and RPGs you usually have more options as in ability to interact with more things and character customization. Minecraft would be classified as an Open-world game. Other games, like basic arcade games like Tetris and Pacman, have fewer options and tend to be less engaging. In a simple arcade game that has no character, such as Tetris, or in a movie or a book, less of you changes. There is less that you can control and interact with in these examples, and less that you can choose about your features.

A reason why we may be wrong is that a very interesting Mario game could be much more engaging than an RPG where you have many options, but maybe with terrible graphics and bad storylines. Our response to that would be to exclude the variables of graphics and storyline. If you could customize your Mario character to have your style of attacks and clothing, and everything was intractable, it would probably be more fun. Another big counterexample would be that movies and books are very engaging, yet you have no options besides pausing the movie, and the pace at which you read the book. Maybe this is because books and movies expose you to experiences you do not have in real life. Maybe the main character of the book is very competitive, and you don’t have that quality. This makes it exciting because you are experiencing something new and are forced to try it out and see if you like it. Some people may think of a basic arcade game such as Tetris to be much more engaging than an RPG. Our theory on this is that these people are used to playing the basic games and are confused by the Open-world and RPG games that have so many options and can be too overwhelming. Some people haven’t gotten used to all the options, so they stick to the simple games that are less confusing and are easier to play.

6. Concluding Remarks

Consciousness as a topic of philosophical inquiry seems to be a natural way to inspire rigorous thinking in middle-schoolers. Children, like the child inside of many adults, are naturally curious about consciousness, and curiosity seems to be a key to rational thinking. For over twenty years, Gopnik and her colleagues have been reporting studies suggesting that babies and children are often forming hypotheses during play, conducting experiments, and doing statistical analysis on their data (1999; forthcoming). They argue that children are like little scientists (indeed, Gopnik prefers to say that scientists are like little children). She has also suggested that babies and children are like little philosophers (2009). These observations might lead us to think that when we struggle to show adults the joys of inquiry, it is not because inquiry is a special skill that must be learned, but rather a natural tendency that we must prevent from being unlearned. Solving puzzles about consciousness is going to require a great creative leap, and creativity comes naturally to children. Encouraging children to think rigorously about consciousness might be our only hope for solving some of these most enduring human problems.

The students in this program were not exposed to philosophical theories. For most sessions, they were simply given thought-experiments as stimuli followed by questions, and taught some philosophical notions along the way (such as hypothesis and counterexample) to facilitate their inquiry. Their own paths of inquiry often mimicked discussions found in the philosophical literature. For the ideas presented this paper the students were simply given an invitation to ask questions of their own, and discuss with the the other students their hypotheses, reasons, and why they might be wrong. You may notice that the hypothesis for question (1), Where is your mind when playing a video game? is in the direction of the view in "The Extended Mind" of Clark and Chalmers (1998). Clark and Chalmers argued that your mind is not confined to the boundaries of your skull, but reaches out into the world it experiences. Likewise, a question similar to (2), Who are you when playing a video game? was explored by Cogburn and Silcox in their book, Philosophy Through Video Games. In their first chapter, called "I, Player: The Puzzle of Personal Identity (MMORPGs and Virtual Communities)," they ask the personal identity question, Is it you who carried out the tasks in the video game? Moreover, the students’ initiative to survey their friends to gather intuitions about the levels of engagement of various games suggests a natural inclination for experimental philosophy of the sort initiated by Weinberg, Nichols, and Stich (2001) and Knobe (2003).

References
Clark, Andy and Chalmers, David. 1998. The extended mind. Analysis 58:10-23.
Cogburn, Jon and Silcox, Mark. 2008. I, player: the puzzle of personal identity (MMORPGs and virtual communities. Philosophy Through Video Games. Ch 1. Routledge.
Dennett, Daniel. 1978. From Brainstorms: Philosophical Essays on Mind and Psychology. Bradford Books.
Gopnik, Alison, Meltzoff, Andre, and Kuhl, Patricia. 1999. The Scientist in the Crib: What Early Learning Tells Us about the Mind. William Morrow and Company.
Gopnik, Alison, and Wellman, Henry. In press. Reconstructing constructivism: causal models, Bayesian learning mechanisms and the theory theory. Psychological Bulletin.
Gopnik, Alison. 2009. The Philosophical Baby: What Children’s Minds Tell us About Truth, Love, and the Meaning of Life. Farrar. Straus, and Giroux.
Jackson, Frank 1986. What Mary didn’t know. Journal of Philosophy 83:291-95.
Knobe, J. 2003. Intentional action in folk psychology: an experimental investigation. Philosophical Psychology 16: 309-24.
Lipman, Matthew. 1974. Harry Stottlemeier’s Discovery. Upper Montclair, NJ: Institute for the Advancement of Philosophy for Children.
Locke, John. 1689/1975, II, xxxii, 15. Essay Concerning Human Understanding, Oxford: Oxford University Press.
Weinberg, Jonathan., Nichols, Shaun., and Stich, Stephen. 2001. Normativity and epistemic intuitions. Philosophical Topics 29: 429-60.


Back to top

The Computer: A Very Short Introduction

Darrel Ince (Oxford, UK: Oxford University Press, 2011). xi + 140 pp. ISBN 978-0-19-958659-2.

Reviewed by Katalin Bimbó, University of Alberta, Edmonton, Canada

The book—as the title indicates—belongs to the "very short introduction" series, which has been published by Oxford University Press for nearly two decades. This imposes certain limitations such as the moderate length of the book, but at the same time, a reader might expect to be able to tackle the text without specific knowledge of computer science. The Computer lives up to this expectation—while it provides a wealth of interesting and useful information.

The book comprises eight chapters, from "The naked computer" to "The next computer" through the small, the ubiquitous, the global, the insecure, the disruptive, and the cloud computers. At first, the chapter titles may seem nothing more than a cleverly constructed sequence of puns, but actually, they reflect the content of the chapters fairly accurately.

The author starts with the general concept of computers and he defines what a computer is. The definition goes as follows.

The description is intended to capture a wide range of computers, and the first chapter already gives a number of concrete examples of computers—from Elliot (a computer of the 1960s) to Deep Blue (the chess victor of the 1990s). Here the author also outlines the themes that he highlights in the later chapters to paint a broad picture of computers and of computing, as well as of their use and usefulness (or harmfulness). The three driving topics are the development of hardware, of software, and of the Internet. These advancements, in turn, caused drastic changes in the understanding of what a computer is and where and how it can be put to use.

Chapter 2, "The small computer" gives an outline of the typical hardware components of a computer. Not only are the base-2 number system and ASCII mentioned, but the author manages to outline the fabrication procedure of a silicon chip, the major types of memory, and various file storage systems. Although the author does not write a history of hardware, he provides sufficiently many details to give an idea of the scale at which computers became smaller and more powerful in the last fifty or so years. These changes lead straightforwardly to the topic of the next chapter.

"The ubiquitous computer" reminds the reader (or demonstrates to him) that computers are everywhere. Beyond miniaturization, which was emphasized in the previous chapter, the ruggedness of chips and the emergence of wireless communications are mentioned as further factors contributing to ubiquity. The author does not shy away from pointing out both the pros and the cons of the potency of ever-present computers: he gives many examples where unobtrusive computers and the data they can generate and transmit are welcome additions, but he also cautions about possible safety and security problems. RFID tags and cell phone apps, when used in large numbers, generate enormous amounts of data—just as certain types of scientific experiments and measurements do. The problem of handling and analyzing huge amounts of data has been nudging researchers toward across-the-board cooperations for decades.

The next chapter, "The global computer," takes up the issue of largeness from the point of view of the amount of computation that has to be performed. The author introduces the class of NP-complete problems, which he colorfully labels as "wicked problems," using two easy-to-understand examples. Problems, like the traveling salesman, are solvable in principle; however, they are too complex for their solution to be feasible to compute in all but very small cases. A practical way to deal with such problems is to give up the requirement of obtaining an exact solution and, instead, settle for finding a "good enough" one. The author mentions as potential approaches to discovering approximate solutions genetic algorithms and swarm computing, as well as variants of AI—before turning to the idea of combining computers. Supercomputers, Beowulf clusters, and grid computing are different ways of amassing the computational capabilities of many processors. In each case, coordination and communication between the processors is required either within a stand-alone computer or over longer distances.

Chapter 5 bears the title "The insecure computer." This chapter would be a very useful read for every computer user because it explains some of the risks of interactions between computers. (Presumably, everybody knows the benefits.) The author provides a list of malware actions (which should appear quite scary to most users) and he gives advice on how an average user (who is not an IT or CS expert) can protect his computer. On a slightly more technical side, elements of cryptography are described (such as symmetric ciphers, public-key cryptography, and SSL). Earlier, the author mentioned the idea of separation, which is a good start (though hardly doable by an average user) toward complete security.

The next chapter, "The disruptive computer," is devoted to some of the economic and social effects of the latest developments in computer technology. The industrial revolution disrupted the wheelwright trade (among others), and similarly, the latest advances in computing replace certain established types of businesses. The most widely known examples come from the entertainment retail industry, where digitized forms of products (from music to movies to books) enabled online distribution and sale, resulting in incredible cost savings. These changes may be perceived as mainly negative; however, the author points out (following Chris Andersen) that a positive outcome of digitization is the wider availability of a greater variety of products. Beyond the commercial aspects, the very same technologies that are used in digitization opened up creative and artistic fields to the masses. Nowadays, anybody can set up a recording studio, create flicks, or take and edit photos—all of which appear to be superior to their 30-40-year-old counterparts.

"The cloud computer" explains the cloud computing paradigm. The author paints a somewhat gloomy picture of computer programs running only on centralized servers. The options, then, seem to be free and inferior software (e.g., a simple word processor or a spreadsheet with basic functionalities) or highly specialized and expensive software (such as an accounting application or a recommendation program for targeted marketing). The author mentions a parallel (following Nicholas Carr) with electricity becoming a utility "flowing" from centralized sources. One should wonder whether a closer analogy can be drawn with the electric motor of which it was once thought that one medium sized per household would suffice. Some widely used instances of cloud computing are the social networking platforms, where the users do not compute much; rather, they distribute content and engage in small-scale (self-)advertising.

The book concludes with an inevitably speculative chapter about the future of computing. The main focus here is on new types of computers, some of which might even depart from the so-called von Neumann (or stored-program) structure. The versatility of modern computers is due (at least at the level of processors) to their capability of inputting programs as data; then, instructions are executed one by one. Neural networks have a different structure, which was originally motivated by ideas about how neurons in the brain form a network. The most radical divergence from the currently existing computers would be quantum computers and DNA computers that are already in sight. They are in their initial phase of development, that is, they are at the stage of the "proof of concept." However, theoretical work on their properties shows that if they become reality, then they will revolutionize computing. It is well-known, for example, that quantum computers would effect a complexity jump.

This little book will be welcomed by anybody who would like to form an overall picture of the current state of computing and some of its future perspectives. The author gives an interesting selection of concrete examples, which enliven the book quite a bit, but he does not overload the text with computer science terminology. References are provided (at the end of the book) for those who would like to pursue further some of the themes. I am sure that philosophers interested in problems related to technology and, in particular, related to computing will find The Computer a thought-provoking book and a gratifying read.


Back to top