Cognition, Learning and Social Change Conference Summary A structured summary of the proceedings of the first conference The purpose of this series of three conferences is to build a bridge between cognitive and social sciences, especially economics. This first conference focused mainly on learning processes, the role of external representations and distributed (social) cognition. The variety of themes that were discussed could be classified under four headings: Internal brain processes and the role of binding; The role of external symbols in cognition; Social aspects of cognition; The role of beliefs in cognition. 1) Cognition inside the brain: Psychology and cognitive science have ignored the economists' rationality hypothesis and focused on cognition as such. The mechanisms of socially distributed cognition systems have become the key to understanding the development of human societies, where learning is primarily a social activity. Still, the human brain is not fundamentally different from primates, only larger. External impulses trigger brain activation and thereby produce "cognitive states". These are not genetically determined. A debate emerged on the meaning of domain specificity of brain activation (theory-of-mind models). While there is agreement that domain specific brain functions exist and that brain architecture must have evolved in the long run (Donald), it is not clear whether there are contemporary variations in domain specificity across cultures and whether that could explain differences in contemporary cognition and its social/institutional expressions. A single "cognitive state" is normally composed of activations and connections between many regions of the brain; the "bindings" between these regions constitute a "cognitive state" (Damasio). While binding is done internally in the brain, external symbols can also induce binding (see below). It is the increasing complexity of these bindings, and of the external symbols that facilitate them, that fuels learning processes.
Cognitive states can be memorized, recalled and modified. What is it in the human mind that allows changes in these states, in the routines of thought? One answer is new information impulses that create new bindings between existing and new states of mind. Another answer could be "drift", the gradual shift of bindings towards new combinations, either because external symbols change the pattern of bindings (see below) or because repeated internal reprocessing creates new bindings. Still, it is obvious that some states of mind are easier to modify than others. The question of the motive for learning was raised several times but not really discussed. Given our information processing constraint, humans are knowledge optimizers, not knowledge maximizers. But what shapes our objective function in that optimizing process? 2) The role of external symbols in cognition The use of symbols is considered to be the major distinction between human and primate cognition (Donald). It has allowed humans to move from episodic to mimetic memory and external symbolic storage, and further to theoretic abstract thinking. The development of language also allowed for meta-representations or inferences between representations, ultimately ending up in heuristic systems that provide "mental shortcuts" (Donald: "theoretical thinking"). External symbols affect cognition in at least two ways (Zhang): they facilitate categorization (synchronic binding) and trigger memory (diachronic binding). This leads to the so-called Whorping hypothesis: the language you speak (the set of external symbols that you use) determines the way you think. Or: the mind's behavior is a function of both the real and the symbolic environment in which it operates. This has several implications: a) Diachronically, cultural history matters: how we have come to contemporary external symbols will affect their interpretation. Path dependency becomes an important factor in cognition. b) Synchronically, cross-cultural differences in external symbols may explain differences in cognition, social interaction and, ultimately, institutions. Zhang provides experimental evidence from mathematical symbols (figures) in different cultures. Is there wider evidence available? The more external storage, the easier cognitive tasks become for a brain with limited information processing capacity. External storage has a leverage effect on human cognitive capacity. It explains why humans have overwhelmingly outpaced primates in resource accumulation and social/institutional complexity.
Furthermore, as the informational complexity of external symbol systems increases, the more efficient they become in information storage and transfer. Cfr: evolution of language symbols: thousands of pictographic Chinese characters, 26 abstract letters in alphabet, 2 symbols in binary computer language. Symbols can be amalgamated into complex sets of symbols (language, stories, models, mathematical formulas, etc.). A coherent set of rules of manipulation (grammar, mathematical rules, logic, software interface) can then be applied to modified outcomes. This cognitive processing happens outside the human mind but produces results that are interpretable by the mind. In short, external symbolic storage (and increasingly, processing) is of key importance in shaping human cognition, both the amount of information that we can process and the way we combine information in different cognitive states (perceptions). While there is apparently little difference between contemporary neuro-physiological states of the brain, there are huge differences between symbolic systems in human societies. These symbolic systems determine our interaction with social and physical environments as well as the institutions (behavioral constraints) that we create to canalize that interaction. The roots of institutions are thus to be sought in cognition. 3) Social learning and role of symbols Symbols can be developed by individuals on their own but are mostly developed in social interaction. In an abstract way, Hutchins defines a lexicon of symbols as a shared view on distinctions (categorizations). Experiments with autistic patients show that the ability to make content-dependant inferences (categorizations that include social context information, "mind reading") is essential in social learning. Sharing events is not a sufficient condition for reading others' minds; shared categorizations are required. Young children have event memory but no intentionality or binding of these events with social context. Shared symbols are more stable and difficult to change than individual symbols; they require coordination among many individuals. Their emergence can possibly be explained through information cascade models (Bikchandani & Hirschleifer; Brian Arthur) or theory of focal points (Sugden) or dissonance based enforcement (cfr. Ash experiment). It was also suggested that a "culture of honor", where individuals are expected to react very strongly and aggressively to any deviation from cultural norms, prevents change. However, these are all lockin models; how do we explain change or lock-out? Possible lock-out mechanisms could be: cognitive drift of categories, Kuran's generational lock-out model, deviant opinions and credibility (see below), watershed models with increasing discrepancy between people's inner beliefs and external expressions (social pressure, central control). This subject was not really explored.
Because of the coordination problem and lock-in nature, shared symbols provide cognitive stability. They liberate scarce brain processing capacity to focus on more urgent tasks. At the same time, they also create inertia in perception and may constitute an impediment to changes in cognitive states (learning). The may also lead to "preference falsification" (Kuran): social pressure prevents people from expressing their real opinion and ultimately stops a social dialogue from taking place (cfr. the Ash experiment). Cognition based on locked-in (sets of) symbols is sometimes perceived as conflicting with utility maximization and rational choice in the neo-classical economic paradigm. However, under the bounded rationality hypothesis (assumes limited human information processing capacity) this need not be the case: it is rational to save scarce information processing capacity for more essential processing tasks and use memorized cognitive states for other tasks. In psychology, dissonance theory and personal construct theory also provide explanations for the relative stability of cognitive states: conflicting information induces dissonance and requires considerable re-computation of many other behavioral routines. 4) The role and dynamics of beliefs Symbols may be amalgamated into complex sets that constitute "stories" or descriptions of perceived reality, at least for individuals who share a similar set of categorizations. Symbols can acquire explanatory characteristics when they condense a complex perceived reality into a simpler mental construction that identifies presumed regularities in an apparently chaotic set of perceptions (Hutchins: a model is a set of regularities within a set of distinctions). For instance, the most primitive form of a model is a "story" that claims to explain certain aspects of observed reality. "Stories" facilitate understanding through binding (categorization) of observations with existing categories in the mind. The regularities are not observed and verified in that case; observations are simply tied to regularities that exist in the mind. Beliefs develop into "knowledge" when independent and repeated verification of regularities is possible (empirically verifiable knowledge). The presumed regularities then become verified regularities. Verification provides no guarantee for its correspondence with "reality"; individuals with different categorization sets may end of with totally different perceptions. Cognitive states remain at the level of beliefs, though they may be "verified" or empirically tested by other individuals with similar cognitive categorizations. Still, moving some aspects of cognition from unverified beliefs to verified knowledge is an important element in societal development. But all societies have their beliefs ("myths"); it may just be a question of choosing "better" myths that enable society to reach a higher level of material resources.
If shared symbols concern comprehensive belief systems (ideologies, religions, that claim an explanation for everything), exclusiveness is necessary: there can not be two or more competing belief systems regarding one subject, one would undermine the credibility of the other. One deviant individual is enough to undermine credibility ("the king is naked" story); strong repression of deviance is thus necessary to maintain comprehensive belief systems (dissonance based enforcement). A debate on the speed of adaptation of behavioral routines followed: Turkey vs Kenya. Is formal institutional change enough to change behavior, or does is require change in the underlying beliefs and norms of behavior? Different views on this subject. Also debate on a particular belief: sharing (¹ shared!) norms. Societies with high variance in resources or weak links between effort and resources, tend to have a higher propensity for sharing resources (Platteau). This is also confirmed by laboratory experiments: participants readily accept unequal distribution of gains if earned through effort or skill. This finding suggests that some behavioral norms are determined by natural environments and not only by cognitive categorizations. Conclusions and next steps The purpose of the conference was to get a better understanding of the origins, underlying cognitive structure, formation, diffusion, transmission and constraints on change in human knowledge and belief systems. How successful has this conference been in providing answers to these questions? We have certainly gained a better insight in the formation and underlying cognitive structures of human knowledge. Discussions related to binding at the level of the brain and categorization at the level of the mind, have contributed to that. Also, the key role of external symbols in diffusion and transmission has been discussed. Furthermore, a productive debate took place on the role of beliefs and the constraints they put on change in human knowledge. Some areas may require further clarification, including the social dynamics of belief systems, the evolution towards more efficient symbolic systems for knowledge transmission and the motives for knowledge acquisition. Altogether, the first conference has thus largely completed its agenda. These accumulated insights in the basic mechanisms of individual and social cognition set the stage for the second conference in this social change project. It will shift attention to the transmission of knowledge in society, both through symbolic means and embodied in hardware devices (goods and services), and will also attempt to bridge the gap between cognition and institutions.
This summary was written up by and filtered through Bertin Martens' cognitive structures, as of 11/1/00.