Scattered Naive Theories: Why the human mind is isomorphic to the Internet Web
[1]Dept of Behavioral Sciences Ben-Gurion University PO Box 653 Beer Sheva 84105 Israel
http://www.bgu.ac.il/beh/david.html
“You mean, said Arthur, you mean you can see into my mind?”
“Yes”, said Marvin. Arthur stared in astonishment. “And.. ?” he said.
“It amazes me how you can manage to live in anything that small.”
The Restaurant at the End of the Universe Douglas Adams 1980
This paper attempts to draw an integrative picture of naive knowledge and understanding. In countless interviews of people about everyday topics (illness, weather, economics, couple formation, old age, dieting, etc.) I was struck by how poorly they (we) can justify their assertions, how hesitant they are and how often they change their mind when challenged. Outside our domains of expertise, our understanding is extremely shallow. Individual judgments are largely made independently from one another, each with a backing that involves only a very small part of our potential, of our store of knowledge and our critical faculties. Coordination, I will claim, take place in Working Memory (WM), and cognitive psychology has long since established how cramped a space this is. Units of explanation and domains of coherence are therefore small. The claim is not that our formulated ideas form an almost haphazard collection of random fragments, though I do think that we tend to be overgenerous in our unreflective assessment of the coherence they exhibit. Some factors lead to coherence, beyond the cognitive efforts of the individual. The domain may be “imported” already structured. This is the case with the many domains that have substantial practical or social importance, and were structured by the society at large. Alternatively, the domain may be one where the conditions for sub-representational learning are fulfilled. The circumstances for this to occur are not well-understood at the present, but enabling factors include frequency of encounter, availability of feedback, and specialized readiness, perhaps an inborn “acquisition device” or preparedness to lean (Elman et al. 1996; Keil 1994). But whenever these conditions are not met, people don’t structure their formulated knowledge spontaneously. In everyday life, that is, most of the time, in free conversations, in interviews, even on topics of interest to them, their ideas tend to be unsystematic and are often incoherent. This incoherence is not usually noticed by those who hold the ideas, and it is only when inconsistent beliefs are brought together to consciousness that an attempt is made to harmonize them.
The following short except from an interview will serve as an opening example:
Do you know what it means to “catch a cold”? Yes. Do you believe there is such thing. Yes. What is it, what happens?. It’s when you go outside on a cold day and are not dressed properly, and it is very cold, then the body reacts by catching a cold. What are additional symptoms? A runny nose, headaches, sometimes backaches. What causes this? Exposure to cold. What exactly happens to the body when this happens? I believe that the coldness affects the body defenses, the body is not strong enough to defend itself against the cold, and then you catch a cold. What causes the cold then? You cannot say it is germs, because it is not germs. The body is busy warming itself and … Wait wait wait, I have an idea: the body fights the coldness, it is busy warming itself and then all kinds of germs come inside the body and cause the cold and throat aches. And what is the difference with a ‘flu? A ‘flu is caused by germs. … Let’s make some order - you said many things. So, you go outside on a cold day, not dressed properly and your hair is wet. What happens? The body is busy warming itself, and then it gets cold, germs attack it, or it just catches a cold. Is this how you always thought about it? No , I didn’t think about the whole process. I am always careful to dress warmly when it is cool at night, but I did not think about why I do so.
People “know” that one is liable to “catch a cold” from exposure to cold. They must have on occasion felt a shiver when out in a cold rain. They also know that illnesses are caused by germs. These pieces of knowledge sat quietly together in our subjects’ mind until they were questioned about just how exposure to the cold may bring about the “cold”. At this point, and then only, did they become aware of the tension between the two beliefs, whereupon they tried to find, on the spot, some way to reconcile the two beliefs, which usually took the form of a facilitation effect: exposure to cold facilitates the invasion. [2]
It is therefore inappropriate to study someone’s naive theory on “what is a cold, and how do you catch one” or on the relation between birth order and personality traits, or on how dieting helps to reduce weight, to mention some topics we studied. Our beliefs on such topics do not form, cognitively, a natural kind. From the cognitivist perspective, the question has not real object, and should be reframed. In the following pages, I will present a number of theses about how our knowledge is acquired, organizedand and accessed, and these will help to bring out how one should think about putative naive theories.
Our knowledge consists in different cognitive sub-systems. In the following sections, we will review several ways in which this saying is true.
The distinction between declarative and procedural knowledge is intuitive and ancient, and received an early formulation in the work of Piaget, who studied sensori-motor and representational knowledge , as well as the transition from the former to the latter [3] (Piaget 1945/51 1974/78; Furth, 1975). The succession of cognitive tools that develop in the course of childhood leave their mark. Vermersch (1979), referring mainly to Piaget’s work, contends that developmental psychology is relevant to understanding the modalities of adults cognition functioning: “ adults display a plurality of cognitive modes of cognitive functioning that may be defined and identified on the basis of the discontinuous character of the development of cognitive instruments.”
Ryle (1949) distinguished “knowing that” and “knowing how”. Under Ryle’s formulation, declarative knowledge can be “declared”, which implies a propositional form and conscious knowledge (Knowlton 1997). The distinction forms the basis of Anderson’s well-known ACT models (Anderson 1983 1993) albeit in modified form. In ACT, procedural knowledge itself is represented in a propositional fashion by so-called “production rules” (Newell 1990; Clark &Toribio 1998), while concepts and productions are connected by webs of associations that determine the probability of their being activated and therefore the likelihood of their affecting ongoing cognitive activity. This departure is fundamental to the model, since the condition for activating a production rule is formulated in terms of the very concepts that form the declarative knowledge.
Sloman (1996a 1996b), in an extensive review of the literature, made the case for a related contrast between associative and rule-based processing. One system is associative because its computations reflect similarity structure and relations of temporal contiguity. The other is rule-based because it operates on symbolic structures that have logical content and variables and because its computations have the properties that are normally assigned to rules. Detailed review of numerous phenomena reported in the literature and in particular the existence of simultaneous contradictory beliefs is marshaled as evidence for separate associative and rule-based systems, the classic example being the Muller-Lyer illusion, where two lines are known to be the same in size yet perceived to be of different sizes.
The distinction between the two knowledge systems underlies the debate in a fairly well-researched substantive field, that of naive or intuitive physics. It has long been known (e.g. McCloskey 1983, Shanon 1976; Piaget & Garcia 1971/1974) that adults knowledge of simple trajectories is systematically flawed. There are two classes of explanations for these findings. The first postulates a general mental model that dictates the form of their errors, such as an Aristotelian or medieval model of mechanics. A problem with this view is that people seem not to be internally consistent (Cooke & Breedin 1994; diSessa 1983; Kaiser et al. 1992; Ranney 1994). The second view is that such knowledge as people have about mechanics consists in a collection of non-verbal cognitive tools (Yates et al. 1988; Cooke & Breedin 1994). These may have evolved independently to handle a variety of phenomena individually, and are therefore situation-specific and not necessarily consistent with one another. The relations between the two types of knowledge are complex. According to Cooke & Breedin (1994), the theory, to the extent that one is requested, is “constructed on the fly”. Keil (1994) similarly contrasts explanation and expertise: “Some forms of expertise might be based on explanatory systems or combinations thereof, whereas others would be based on our fallback mechanisms of association and automatization.” Explanation, he goes on to claim, would rely on a small number of construal systems, while expertise would be constituted of a multitude of tiny systems.
While these dichotomies of procedural/declarative or associative/rule-based are convenient, they collapse several distinctions (Gigerenzer & Regier 1996). Procedural knowledge itself can take several different forms. It may rely on individual traces of many exemplars and past encounters, as in instance theory (Logan 1988), Hintzman’s (1986) Minerva model (see also Aha 1998) or the sedimentation of many encounters (as in neural nets or classical learning models).The variety of ways of representing knowledge constitutes a fundamental feature of our cognitive makeup (Shanks 1997). Congenial views have also been arrived at in the brain systems literature, where declarative knowledge has been operationalized as that which amnesic patients cannot learn, or only with great difficulty (Knowlton 1997, Shimamura 1986). The prevailing view is that its converse, “non-declarative knowledge”, does not correspond to a single system, but covers multiple implicit processing systems that may have different properties (Squire, Knowlton, & Musen 1993). It turns out that an impressive range of learning phenomena can be demonstrated in amnesics: skills, priming effects, artificial grammars, fuzzy categories and exemplar based knowledge are all preserved which implies, in this perspective, that these manifestations of learning do not rely on declarative knowledge (Knowlton 1997).
The following illustration relates to a topic to which our subjects (young men and women) had given extensive thought: what makes for a stable couple (Sher 1999). The topic is itself if of central and personal importance to young adults (aged 20-30), it forms the focus of frequent gossiping, the background of many movies and books. If a topic like this was not systematized in a theory, what will?
The interview included several phases. (1) List 3 women and 3 men you know well, by their first names. Would Male2 and Female1 form a stable couple?, would Female2 with Male3? (2) Explain your responses. (3) Suppose you were a professional matchmaker, what would you be looking for ? (4) Please list several couples you know well. (5) Do your principles fit these couples?
When asked whether, say, Ariel and Yael would form a stable couple, subjects pondered for a while, then produced confidently an answer, though as a rule they had the greatest difficulty in justifying it. There is no telling what exactly went on in their mind, but it certainly did not consist in checking whether a certain known formula was true of the proposed union. In the next phase of the experiment, they were asked in effect what are the conditions for forming a stable union. Subjects almost always claimed at the outset that resemblance is a first requirement (in personality, interests, phase in life, background, even physical resemblance). After a while, they took this back, and became puzzled, for that theory rarely fits the couples they know. Some then proceeded to say that some properties must be similar, others need not, but find it hard to specify this further. Some also introduce, later on, the importance of complementary properties, such that partners balance each other. “He is quiet, she is outgoing. He gives her peace; she forces him to get out of himself. “ And many subjects conclude that love is irrational: it is a special kind of “chemistry”, it has to “click” – which is an avowal of failure.
Clearly, more than one system involving knowledge about stable unions is at work here. Subjects feel fairly confident in deciding, intuitively, non-declaratively, whether a proposed match could be successful, they are personally acquainted with existing couples, exemplars that could constitute an empirical basis for induction, and they have some principled, declarative notions about what to looking for in a match. But as soon as these systems were confronted, confusion prevailed. An array of independent implicit systems may be perfect to enable functioning in different circumstances, but scarcely supports the development of coherent, comprehensive naive theories.
Our knowledge relies on diverse systems. In the cognitivist sense, this means that different representational formats coexist in our cognitive system. In the brain systems perspective, different brain structures are involved in representing knowledge. The formation of knowledge in each and the interactions between the systems, constitute one source of discrepancy between the constituents of knowledge.
(a) Let us now focus on propositional knowledge. We will see that the ways that knowledge is formed leads to inconsistencies. We may distinguish two modes of propositional system formation, according to which of its two aspects, “propositional” and “system”, precedes the other.
The first mode might be called, following Piaget, reflective abstraction (1978) or “cognitive phenocopy” (Piaget 1974). In this mode, a given knowledge domain in the propositional system does not form as such initially, but is based on a pre-existing associative or sensori-motor. Knowledge of non-propositional systems occurs in extracting and exploiting regularities in the environment. That knowledge is consequently embedded in an enactive system that generates predictions or behavior, and therefore at first restricted to the specific context in which it was acquired. For example, in the case of naive physics discussed above, people may develop separate systems for predicting trajectories of curve balls and catching them, that of bombs dropping from airplanes, or of objects rolling off a steep cliff. The systems involved in throwing objects so as to attain a given aim need not be the same as that involved in catching an object or in predicting a trajectory from the side. Indeed, they rely on different system according to whether they have to predict a motion or judge its naturalness. Kaiser et al. (1992) report that when presented with animations of certain motion problems, people judge their own predictions to be unnatural and select natural motions as appearing correct.
Reflective abstraction involves the twin aspects of change of representational type, and restructuring at the higher level, possibly by way of various implicit levels (Dienes & Perner 1999; Piaget 1978 1985). Such a transformation to an explicit level has of course important consequences. If we consider it from the perspective of the declarative subject, it makes the system “cognitively penetrable”, paving the way for the subject to become aware of the structure of his or her beliefs, and in particular of contradictions. The powerful range of linguistic and logical reasoning tools becomes in principle applicable. For Piaget, sensori-motor knowledge is situated (Furth 1975; Bic& Campbell 1989), so that the transition does not merely consist in extracting information implicit in an enactive system, but also in internalizing the environment separately from the subject’s reaction to it. The difficult in doing this is the source of cognitive egocentrism. We might add that if several systems are subject to reflective abstraction, the resulting egocentric representations need not readily be coordinated.
Karmiloff-Smith (1992), a developmental psychologist, studied and elaborated this “bottom-up” direction, from systems of implicit knowledge embedded in procedures, to a level of explicitly defined knowledge not available to verbal report, and finally a level available to verbal report. On her view view, “Both innately specified knowledge and newly acquired knowledge are initially procedurally encoded, activated as a response to external stimuli. The procedure as a whole is available to other operators, but its component parts are not” (Karmiloff-Smith ,1993) [4]. Of course, reflective abstraction need not occur, or may take place only very partially, and knowledge may remain largely at the implicit level, a fact increasingly emphasizes by students of culture (Bourdieu 1990; Bloch 1998; D'Andrade 1995; Strauss & Quinn 1998) and of organizations (Baumard 1996/1999; Sparrow 1998).
Procedural, sensori-motor, enactive systems develop independently to solve different pragmatic requirements. There is therefore every reason to expect coordination between them to be difficult, and extracting the propositional content implicit in them yields different theories. This then is another source of disparity.
(b) The second mode of propositional system formation goes from the acquisition of individual propositions to their coordination into a system. Individual propositions can be acquired before a system is fully ready to accept them. An early proponent of this approach in education was Vigotsky, who distinguished “spontaneous” and “scientific” concepts (Vigotsky 1934/1986): it is well to teach rather remote topics to children, he taught, since this will enable them to form a good conceptual system unencumbered by the associations of everyday concepts.
Wilkes (1997) reviewed the literature on “accumulative memory”, his term for the ability to accumulate information about the world and maintaining control over the open-ended and constantly changing inventory of memory records. His summary is that : “(1) memory records can be organised to avoid interference (not always very successfully) (2) records may be integrated over time (not always very exhaustively) and (3) records may be updated (not always very extensively). We may do all of theses things some of the time, but there are many occasions when our efforts are, at best, perfunctory.” (p. 172) Even these conclusions are probably too optimistic, since most of work reviewed is concerned with laboratory experiments, where context effects are minimized. Wilkes conclusion is inescapable: “accumulative memory has evolved to make information available when needed, not to put it on trial. There are other ways and times for doing this” (p.173).
These two ways in which propositional systems form, reflective abstraction of non-propositional systems and coordination of individually acquired propositions, lead to the same outcome: uncoordinated knowledge.
The need for coordinating propositional knowledge is less than would seem from the previous analysis, and there no proliferation of wholly unrelated individual propositions because propositions we come to accept do not emerge individually. A relevant distinction is developed by Sperber (1990 1997), a cognitive anthropologist, who discusses two systems: intuitive and reflective beliefs. Intuitive beliefs are derived, or derivable, from perception, and are limited to basic concepts referring to perceptually identifiable phenomena and to innately preformed unanalyzed abstract concepts [5]. More important to our present point are “reflective beliefs”. A reflective belief is a proposition held by someone to be true on indirect grounds, such as trust in someone who asserts that it is true. To hold a reflective belief, one need not even have a well-developed conceptual system that gives it meaning. One way or another, the belief in which they are embedded validates them.
While it may be true that even individual, unconnected propositions may disseminate widely in a society , (Sperber strikingly uses the term epidemiology in this context), beliefs acquired from others will usually partake of systems whose properties are ultimately the reason why society or individuals adhere to them. Socialization into a culture or social group provides therefore a great deal of such coherence (see e.g. Breakwell & Canter 1993; Carey & Spelke 1994; Doise 1989; Duveen & Lloyd 1990).
But saying that there are coherent systems in culture or education does not mean that those systems cohere with one another. On the contrary, several such systems will typically co-exist. Here is an illustration, from a study of parents’ understanding of the epilepsy that afflicted their child.
One of our informants was a Bedouin mother who attributed her child’s epilepsy variously to the bad eye (an evil neighbor living nearby); a spirit that entered the child while she was lying down; it was her own fault for having eaten lemons and salt during her pregnancy; and the illness could also be traced to the child itself being such a fine, delicate, beautiful child (a designated victim). A second informant was an orthodox Jewish father who said that: illness runs in the family, yet he worried that his son was struck because he does not study the Talmud enough; he too attributed the condition to the bad eye, and also to a sudden fear felt by the child upon seeing foreign workers.
These explanations are not necessarily incompatible:some are predispositions, other pertain to moral retribution, course of illness or episode triggers. The point here is that all were offered as explanations, and no attempt was made to coordinate between them. As Kleinman (1980) noted, “illness is always a cultural construction” and when there are several explanations, the patient’s health-seeking behavior is characterized by simultaneous resort to several sources of relief: physician, dervish, rabbi, etc.
Here are some more examples in the medical domain. Quintero & Nichter (1996) who studied understanding of addiction found five expert discourses on addiction (the physiological perspective, the cultural constructivist model, the psychological model, the political economy model, and the disease model) and some additional ones amongst the lay population. Similar findings were presented by Zani (1993) in her study of mental illness, by Leiser, Doitsch and Meyer (1996) who studied the many sources of beliefs about everyday high fever in mothers, and by Stainton-Rogers (1991) who identified seven attitudes to health and illness. Mulhall (1996), analyzing the “myth of stress in nursing and medicine”, shows in detail how the concept of stress, besides explaining misfortune, has also an ideological and a political function. “The lay and professional concepts of stress blur into each other and are mutually re-enforcing. Stress in both area is taken as a real concept, which does however vary widely.” Since the concept is useful but used for such widely divergent functions, it is understandable that it is “stamped with a hallmark of authority/ambiguity”.
A final factor contributing to the heterogeneity of our knowledge is our mental endowment at birth . Many investigators have come to ascribe to the new-born human infant a preparedness to develop specific cognitive stances, each characterized by a typical domain of application, an ontology, and accounted for by its evolutionary relevance (Gardner 1983, Hirschfeld & Gelman 1994; Carey & Gelman 1991; Keil 1990 1994; Wellman & Gelman 1992; Spelke et al. 1992; Rosser 1994; Wellman, & Inagaki 1997). Each such stance represents a particular appropriate way of looking at aspects of the world, but when two such apply to the sobject, they cannot readily be coordinated. Consider for example the psychological and the physical stance, applied to the mind, resulting in vexing problems such as the relation of body and psyche (how can pills such as anti-depressant affect mood? Where are thoughts localized, in the head?) or the tension between the moral and the instrumental views of the economic world (Leiser, Sevon, & Levy 1990; Williamson &Wearing 1996; Watson 1996).
Another source of inconsistency may perhaps be found in the conceptions that develop successively in the same individual. Developmentalists have not yet come to a definitive conclusion as to how new conceptualizations relate to previous ones, but if the previous ones are conserved alongside the new one, an additional inconsistency may result. Consider understanding of biology, which is known to evolve in normal humans from a simple collection of individual biological facts, into a functional theory. This implies a reanalysis of animals as living things with a universal life cycle and the role of bodily parts and function in maintaining life. From a comparison of normal children with children afflicted with Williams Syndrome, Johnson and Carey (1998) were able to show that the shift to the functional view goes beyond the acquisition of new knowledge formulated over an existing conceptual base (which is accessible to Williams Syndrome children). It involves genuine conceptual change. But what happens to the superseded conceptualization? When the two do not conflict, both may continue to exist. According to Chi (1992), however, even when there is a conflict, the older conceptual system is not transformed into the new one (as some would have it). The new conceptual system grows alongside the first, and dominates but does not eradicate the older one. If the older conception is not obliterated, intellectual history leaves its mark on later functioning.
As an anecdotal illustration, the philosopher Gaston Bachelard (1940) defined « epistemological profiles » of concepts, a notion that applies to a given individual, at a given stage of his or her intellectual history. For himself, he tells us, the concept « mass » is part naive realism, part positivist, strongly classic rationalist, and so on. The importance of the naive realist and classic rationalist dimensions in his mental makeup are a consequence of his specific life history: before he became a celebrated philosopher, Bachelard was a postal office clerk and had to weigh letters and parcels, and later spent several years as a high school science teacher. Similarly, if we consider our own notions of monetary value, we may perhaps recognize both a mature conception (the value of an object is determined by the intersection of the demand and supply curves) and an older, absolute one (every object has its real, intrinsic value – whoever takes more is overcharging).
In the previous section, we saw that our knowledge is based on many sub-systems, but from different types and from different causes: there are the declarative and several non-declarative modes of knowledge, there are cultural systems, and inborn content-related cognitive stances. Different tasks engage different cognitive structures and, since these were never coordinated, their respective output is uncoordinated. The following illustration brings this out:
Subjects were interviewed about the relation between birth order and personality traits. The procedure involved the following steps: (1) Subjects were given a pile of index cards and asked to write down properties (such as vain, social, clever, dominant, etc. ) and place them on one of three piles, labeled Firstborn, Middle Child, and Youngest Child. (2) They were then given a prepared set of cards with additional traits, and asked to distribute or discard each card, then (3) requested to justify their choices. (4) In a fourth step, they wrote down a list of people they know who are firstborns. (5) Finally, they had to say whether the properties they listed for Firstborns actually fit their chosen instances, and discuss discrepancies.
Subjects were very rapid in attributing the traits to the three positions (steps 1 and 2), but tended to find it very difficult to justify their choice (3). Indeed, they often retreated, were puzzled, and sometimes looked as though they were trying to solve a poser prepared by someone else. Further, there were little common explanatory principles underlying the justifications they provided across trait and across piles. There were also no special relations between the instances provided (4) and the traits attributed, except usually for one chosen example (“my big brother, my youngest daughter”). When we asked more pointed questions confronting the various steps, we elicited the whole range of reactions betraying perplexity: avoidance tactics, hesitations, creative generation, retraction, and extensive use of “non-explanatory” explanations. We counted how many explanations had actually some contents, as opposed to circular, vacuous explanations. Overall, there were 68 causal accounts, against 95 vacuous explanations.
Writing properties was fast, associative, non-critical; justifying them involved a range of independent explanations, but no overall explanatory system. There are many vignettes, such as that the middle child is “sandwiched” between older and younger sibling, the youngest is spoiled by doting parents, the eldest has younger sibs and parents give him or her more responsibilities. The subject then derives personality traits from these (the eldest is therefore more responsible, the youngest a self-centered little brat). The inductive base, when it is used, is as slender as can be: there is usually just one paradigmatic case (all firstborns are like my older brother), and no spontaneous reference to any other cases. When the list of nagenerated in the last step is directly examined, it is typically found that it does not fit the pattern. People can and will discuss the relations between birth order and personality without having any systematic account of the link between the two, and propose such links on the basis of an uncoordinated mix of associations, homely truths, episodic memories and folk-psychological principles.
I submit that there is no spontaneous, unconscious mechanism responsible for scouring Long Term Memory (LTM), finding mutually relevant items and coordinating them. We referred earlier to Wilkes’ (1997) discussion of “accumulative memory”, from which it was concluded that the dynamics of cumulative knowledge (Leiser 1997) cannot lead to a spontaneous coordination between knowledge sources and modes. Hence coordination, to the extent it takes place, must involve Working Memory (WM).
This is also the conclusion that emerges from the work of Baumard (1996/1999). Baumard studied a group of professionals specializing in collecting and interpreting intelligence and analyzed their “exemplary management of knowledge”. That knowledge itself consists mostly of small bits of information whose implications are ordinarily not clear cut. The work of those professionals, on a given topic, may be broken down into three clearly marked phases. In the first, the objective is to accumulate impressions. The second phase is one of deliberate learning, concentrating on such avenues and topics that seem to emerge from the first. This remains intuitive, the professional deliberately refrains from forming hypotheses, but searches for knowledge that may be relevant to impressions gleaned during the first phase. In a third phase, he becomes aware of a “complex web of convergences” that he then attempts to explicate. This is hardly a case of naive theory (remember that Baumard’s subjects were highly professional information specialists). But even here, it is seen that knowledge accumulates in LTM, while the patterns that form are deliberately exploited, first to formulate and pursue promising lines of enquiry, and later to integrate, interpret and explicate the accumulated data.
To pursue this claim, let us remember the striking (to them too) lack of consistency exhibited by naiadults interviewed on everyday topics, as in our opening example where we asked adults about the common cold. People believe that exposure to cold, damp weather may cause a cold (Helman 1978). This knowledge is presumably acquired from parents’ warnings about going outside in the winder without proper attire, especially with wet hair, etc. They also must have on occasion experienced a shiver when out in a cold rain. Further, people have been exposed extensively to the medical model, and also know that illnesses are caused by germs of some sort. Only when questioned pointedly by the experimenter did they realize the tension between their beliefs. Sometimes, the conflict was expressed by long pauses, at others the subjects were quite explicit: “Hmm … I never thought about this … wait a minute, I have an idea, I can tell you what I think …”. Awareness of the conflict and its occasional reconciliation awaited our questioning. Conversely, since most such contradictions are not brought into the open, they remain latent without any mechanism to identify them.
Coordination must take the form of a series of searches in LTM, on the basis of retrieval cues generated in Working Memory. However, this search may not be very effectual. According to the “encoding specificity principle”, information acquired in a specific context tends to be retrieved better in a similar context than otherwise (Tulving & Osler 1968; Godden & Baddeley 1975). Moreover, retrieval cues tend to be superficial only (Brown 1989; Gentner, Rattermann, & Forbus 1993; Gentner & Toupin 1986; Holyoak & Koh 1987; Holyoak, Novick, & Meltz 1994; Gick & Holyoak 1980; Morton, Hammersley & Bekerian 1985).
We will discuss in a later section the difficulties and limitations of resolving inconsistencies, once they are identified, but will first present an analogy. To summarize the argument so far. Since our mental makeup at birth and the mechanisms and conditions of knowledge acquisition generate separate systems, and since there are no general mechanisms responsible for spotting and improving coherence in LTM behind the scene, as it were, coherence will be local at best, and attempts to harmonize knowledge will be based on probes originating from the contents of WM (Ranney 1995). Every period finds its own analogies (Gigerenzer & Goldstein 1996; Marshall 1977), and the preceding summary suggests a novel –if predictable –one to the human mind: the World Wide Web. Specifically, I shall claim that naive thinkers using their LTM are akin to internet users trying to understand a new topic on the basis of a www search.
We have a vast store of distributed information, and several search “portals” engines and agents, which allow us to frame search cues and define searches. These may all be set to work in parallel. The result of such a search “swarm” is a set of pointers to various contents that we can bring to the fore and study. The material itself is highly variable. Some may be episodic, an anecdotal piece of information, some might be sites devoted to one aspect of the subject at hand, or to some other subject that is obliquely related to the topic of the query. Some search engines rely on some a indexing that takes place before the search, others are based on surface features. The reliability of the various returns is also variable. Many of the sites found will themselves have further links to other sites, and so on recursively, while some may be permanently or temporarily unavailable. The person who initiated the search may decide to pursue one line of search, and ignore others or choose to integrate information assembled from various hits. After there are some returns and the user has begun working on them, the search engines may deliver new pointers that may then be ignored or stored and perhaps incorporated in the emerging understanding. Managing the hits is a complex skill.
The history of the Internet explains why it is so apposite a metaphor. Information always accrued in numerous locations in isolation from one another in terms of purposes, methods, ontology and so forth. With the advent of the world-wide web and the consequent progress is availability, it has become more natural to consider all that information as belonging by right to a single subject (humanity? the “noosphere”?) so that its fragmentation, incoherence and lack of interconnectivity came to be seen as defects that can and should be remedied. This perceived need to integrate information distributed in the net has spawned research efforts at I3, “Intelligent Information Integration” [6] .The mission of these projects includes handling heterogeneity of all kinds (semantic, schematic, vocabulary, data), resolving inconsistencies, developing languages for information integration and techniques for intelligent information retrieval from Web sources: finding, extracting and synthesizing it.
This mission, it turns out, is extremely difficult, for reasons directly related to what precedes. Information developed in unrelated projects have different pragmatic aims and rely on different conceptual approaches. The principal source of difficulty is in handling incommensurable statements. A typical approach to I3 is taken by the Context Interchange Project (COINS) [7] at the Sloan School of Management, MIT, which favors a declarative approach. This requires that interoperating sources and receivers describe unambiguously the assumptions which they make in the routine representation and interpretation of data. However, meaningful integration requires a conceptualization of the underlying domains which are shared among the component systems. These shared domain models are referred to as the underlying shared ontologies. [8] Data bases must be built in accordance with the ontology, if later coordination is to be possible. But when the data bases and their ontologies have evolved independently, there arises the problem of integrating existing ontologies. And this, Uschold & King (1995) conclude in their survey, is very hard.
Returning now to naive theories, propositions are couched in a language, partake of a culture, belong to particular domains or approaches (medicine, psychology, science, religion, ethics) and these are bound up different ontologies. Any language game must possess its own, but separate games need not mesh.
Moscovici (1990) put it well:”… ordinary knowledge, common sense if you prefer, is really something other than an ‘expertise’ in everyday thinking. It implies a combination … of very different types of thought and of information. It is thus distinguished from specialized or expert knowledge which, to the contrary, attempts to follow a single type of thought and to deal with a single category of information.”
Knowledge typically remains fragmented (diSessa 1997; diSessa & Sherin, in press) and it is only when challenged that people will try to combine the fragments to construct an overall account. However, success is certainly not guaranteed, as there is no known method to combine disparate conceptualizations. For instance Biswas et al. (1997) describe their subjects’ thinking about electricity as follows: “Students are active problem solvers who try to capitalize on whatever knowledge they have at their disposal. This includes qualitative knowledge about relationships, practical facts, mathematical formulas, and metaphors. Students have difficulty integrating their different representations into a mutually constraining understanding. Consequently, even though students switch between representations, they do not always make bridges between the different representations.”
Building an explanation is often done on the fly, with all it implies in terms of hesitation, instability, and dependence on context effects. Let us return briefly to our metaphor of the web user grappling with the bountiful output of search engines. Skill at framing retrieval cues and at managing the resulting flow of information can yield quite different performances, even though different users are exploiting the same net. We found it useful to group our subject’s approach to our questioning by the following typology of answering styles: As, Coherents , and Flexibles. These are really ideal types, but the typology clarifies the opposing requirements subjects are trying to satisfy.
Associatives rely on personal experience, simple superficial analogies, and various information tidbits. They then get stuck, and begin to collect further information either in parallel or successively. If in parallel, sets of unrelated information pile up; their discourse consists in simple sentences and juxtapositions. If successively, one association evokes the next one. Syntactically, they produce few complex sentences; each question raises one set of responses, not necessarily connected to the next; they tend to be unaware of contradictions, and to resort to evasive tactics when confronted.
Coherents are more deliberate: they retrieve (or only report) information more or less coherent with their earlier statements, which form the main retrieval clue. They remain aware throughout of what they claimed, but also of outstanding problems or gaps. Syntactically, their speech is more abstract, general, conceptual; the relation between sentences are clearly coordinated. A last feature: they are very rarely found.
Flexibles raise many associations, examine them critically to see what may be used, attempt various avenues, and frequently retreat from earlier statements. Their retrieval cues define a wide search beam, and the Flexibles attempt to coordinate the results on the spot, without lasting commitment to any one explanation.
All three styles apply unmodified to our metaphor of search on the internet web. Our questions force our subjects, for the first time, to collect what they have in LTM on a given topic, and to actually do what availability of language representation makes possible: confrontation of the different sources of knowledge. Subjects may be more or less nimble at the task. The main conclusion, however, is that they didn’t engage in it before that time.
These guided, deliberate attempts to be exhaustive are necessary because, in the absence of an integrated knowledge, there is no way to tell whether some relevant information was overlooked, solely by looking at what information was used in reaching a decision. Fodor (1983) based his provocative theorem on the “impossibility of cognitive science” on the observation that everything can in principle be relevant to everything else. This is indeed true, but only in principle. And the advance of psychology over philosophy comes precisely in worrying about what information is in fact brought to bear on specific problems. To answer a question, the subjects need to marshal such information they have. To this end, they send retrieval cues whose returns give rise to a subjective Feeling of Knowing (FOK). That FOK is unreliable. It is based on the size of the “echo” evoked by the retrieval cues, and as Koriat observed in the context of specific memory retrieval: “Although such clues may not be articulated enough to support an analytic, calculated inference, they can still act en mass to produce the subjective feeling that the target is there and will be recalled or recognized in the future.” (Koriat 1995). This feeling seems to be based on the overall familiarity of the retrieval cue (Metcalfe & Shimamura 1994; Reder & Ritter 1992), as well as on partial information retrieved from memory (Koriat 1994). Neither of these processes seems to involve access to discrete properties of the items being processed (Markman & Dietrich 1998).
When the subject experiences a FOK based on his or her search cues, all he or she knows is how much material was activated by that first cue. There may be highly relevant material available, but for the moment not accessed, and nothing is known to the subject about the quality and nature of the elements responsible for the size of the echo.
In the course of a study we made on students’ understanding of dieting, we asked our subjects for a self-appraisal of the extent of their knowledge. This appraisal was done twice, once before the interview, that included a variety on questions on how to lose weight and why various methods are more or less effective. Strikingly, most subjects changed their appraisal. Some did this in the direction of an increase. Here are some examples of justifications: “I changed my mind because I figured ‘what do I know about dieting, I never followed a diet in my whole life’ but then suddenly I realized there is a lot I do know, from TV , from grade four. I suddenly remembered. While I was thinking I remembered things that people talk about and that I heard at home, at school, in the army, the newspapers, and I think that I understand a little better than I thought before”. “Apparently it all was asleep in my memory, and I did not realize how much I knew” Others decreased their appraisal: realized I was not able to organize all this stuff, to define them properly, and there suddenly are a million contradictions that I never thought about.” Specifically, girls who reported being engaged a lot in dieting ended with lower scores than before: “all my life I am into dieting, so I figured I knew a lot”
Harmonization of declarative knowledge must involve the contents of Working Memory, and we saw that most inconsistencies will not be noticed, simply because the relevant information will no be brought together in WM. Search for incompatible information does not take place spontaneously. Once information is present together in WM, there are two classes of mechanism that may serve to coordinated the pieces, corresponding to two classes of processes: automatic and parallel, on the one hand, and controlled and serial, on the other [9]. However, since there are strong limits on the number of propositions that can be considered together, large-scale integration is usually beyond our cognitive means.
a. Parallel constraint satisfaction models
In these systems, the degree of coherence of any two propositions or attitudes is known, and the system, starting from a given level of support for each proposition, attempts to adjust the level of belief in each proposition so to obtain a maximally coherent system. Such a system is not fully conscious, though it relies on the concurrent presence in working memory of the set of propositions to be co-adapted. Its output is closed, in that no new propositions enters the set once the process is started, and its output is a novel set of degrees of belief in those propositions. Some beliefs are rejected, other are upheld. Read and Miller (1993) observe in this context that evaluation of coherence is only local: The greater the degree of attention paid to a node, the greater the amount of activation it can send to other nodes (Thagard 1989; 1992; Shultz & Lepper 1992; Holyoak & Simon 1999; Read & Miller 1993; Ranney & Schank 1998).
Holyoak & Simon (1999), for instance, asked their subjects to rate their support for a range of seemingly unrelated propositions and found at first, as expected, no particular correlations between these judgments. When they presented these propositions a, but this time in the context of a trial so that each of the propositions either supports or weakens the case of one of the parties in the trial, it was found that subjects move towards coherence. If a subject judged in favor of one party, she also tended to agree to all of the propositions that support that verdict. A kind of backwards reasoning is involved, from the conclusion (the verdict) to the antecedents, so as to obtain a maximally coherent position. Such a system is not fully conscious, but relies on the concurrent presence in working memory of the set of propositions to be coadapted. Its output is closed, in that no new propositions enters the set once the process is started, and its output is a novel set of degrees of belief in those propositions. Some beliefs are rejected, other are upheld.
A broadly similar parallel-constraint satisfaction model was also proposed (Kunda & Thagard 1996) to accounts for the information flow between stereotypes, behaviors, and traits, and how multiple stereotypes jointly affect impressions. Such constraint satisfaction networks are relatively automatic processes that require little conscious processes.
An apparent objection could come from the work by McGuire (1990), who claims that when a change is directly induced in one part of a belief system, it produces predictable adjustments in other parts. For instance, a core events desirability will be adjusted to its likelihood and therefore to its pattern of salient antecedents. McGuire studies thought systems (the term is taken to include attitudes, cognitions and beliefs) of “moderate complexity – consisting of a dozen or so “topics of awareness”—usually a core event that might befall the person, and its salient antecedents and consequences. He manipulates the desirability or likelihood judgments of a core event, and studies the patterns of salient antecedents and consequences evoked by mention of the core event. While McGuire concludes that thoughts are “organized into connected and coherent minisystems”, the pattern of evocation does not imply anything about the prior organization of the topics into such a minisystem. Actually, his findings only show that desirability and likelihood affect the search differentially when evocation takes place.
The second way to integration consists in taking a set of incompletely coordinated propositions, and constructing an account that would encompass them all as far as possible. Deliberation is a … deliberate, controlled, serial activity consisting in the confrontation of propositions, searching for evidence to decide what to believe, the use of various cognitive moves to remove seeming contradictions (distinctions, dissociations, explanation, auxiliary hypotheses, exceptions etc.). Its output – if successful – is an “account” that may involve many propositions beyond those of the starting set: propositions retrieved or based on contents of the LTM, and novel constructions, created to remove existing inconsistencies. Finding individual explanations is a task we are rather skilled in, consisting as it does in searching an explanation in our store (Abelson & Lalljee 1988; Pennington & Hastie 1988) or constructing a new one, on the basis of preexisting knowledge and patterns of cooccurrence. However, coordinating explanations is not an activity one spontaneously engage in: one is happy with finding the individual explanation, whereas building a wider explanatory theory is not a natural activity.
A special cognitive task is explanation. Providing a complete explanatory system is not a task one usually engages in. People are happy with the most perfunctory or vacuous explanation: aspirin has headache curing properties, just as in Molière’s play its “dormitive” (sleep inducting) powers explain why poppy makes you sleepy. Knowing just the relation is not useless: it does predict that taking aspirin will help you with your headache; when you have a headache, you know what to do. But a narrowly pragmatic focus may not always be enough, and success is sometimes served by getting a grip on a causal explanation.
Toulmin (1958) developed a theory of scientific argument, based on the roles that propositions play within an argument. A claim is supported by data, and the warrant is the general rule by virtue of which the data are a support for the claim; a backing is a reason for accepting a warrant. It may be a recursive justification (treating the original warrant as a claim to be justified), or it may be an appeal to authority or some other reason for belief. Thagard (1998), who analyses the relations between patterns of correlations, causes and mechanism with the view of explaining disease, in his chosen example, offers a broadly similar theory. Patterns of relative sizes of correlation may help us to formulate causal dependencies (Cheng 1997; Glymour 1998) while conversely, a postulated mechanism and pattern of causal dependency helps us focus on the relevant relative correlations. However, in all such cases, the number of propositions involved in establishing a causal explanation remains very small, as it has to stay within the narrow confines of what we can cognitively handle.
If the topic at hand is important to the thinker, there will be a corresponding set of propositions such that several of its members will frequently activated concurrently, in various combinations. If she coordinates a subset every time, little by little the whole may well become coordinated. However, this is a process that would take a long time, and there is no guarantee that it will converge: it may also result in oscillation between various positions. In any event, it will usually happen in the mind of serious scientists in their chosen field , not in everyday activity. This fits in with the conclusion reached by Carey and Spelke (1994): major theory changes occur in the mind of sor other specialists. While they can then be imparted to individuals, it is by a process of education.
This view would be opposed by many, especially cognitive developmentalists, whose very domain is focused on the elaboration of novel theories by their subjects. For Karmiloff-Smith (1993) “The child is a spontaneous theory-builder (about language, physics, etc.)” Similar views are held by (Gopnik & Wellman 1992; Vosniadou & Brewer 1992; Carey 1994,1991). There is no gainsaying that theories do develop in childhood and there is no doubt that their cognivite competences develop with age. But regarding the process itself, nothing precludes its taking place in the working memory of children too. Consider the study by Vosniadou and Brewer(1992; building on the work by Nussbaum & Sharoni-Dagan; 1983) which showed that individual children are actually quite consistent in their distorted and varied understanding of the shape of the earth. Briefly, children are faced with two apparently incompatible sets of data about the earth: it appears flat, yet they receive information about it being spherical. Faced with this contradiction, children develop special models that somehow reconcile the two views. For instance, some claim that the earth is a disk (i.e., both flat and round), see themselves as on standing on a flat surface inside a hollow sphere, or differentiate between the flat surface they stand on and the celestial object called planet earth. While Vosniadou and Brewer demonstrate how far the children are consistent in holding on to a given view, they too recognize that for some (perhaps the more outlandish models) the child constructs the model “on the fly”, when confronted by the experimenters’ questions. Certainly in working with children on Piagetian tasks , it is extremely common to see them resolving the contradiction there and then, after their initial puzzlement.
Children’s resolution of such contradictions may well be based on conscious deliberation too. Consider for instance a study by Mazens and Lautrey (1999) who studied children’s understanding of sound, and more specifically whether sound could traverse solids. At first, their answers seemed incoherent: sound was believed to pass through solids or not, without any apparent system. study revealed the logic of their judgments. Sound can pass through solids provided one out of two conditions is fulfilled: if the sound is stronger than the material, or if there are holes in the material. Further, children have fairly systematic notions about the relative strength of sounds and various materials, and as to whether or not they have holes in them. The theory involved attributed to the sound properties of materials: substantiality, permanence, weight, etc. One can see how children arrived at this theory. On the one hand, knowledge about complex domains relies on the accumulation of information about the frequency statistics of the stimuli and respond to new events on the basis of their featural overlap with stored representations. They had to summarize from past experience when sound did traverse an obstacle and when it was blocked or muffled. This knowledge is collected, but need not be verbalized or ever consciously accessed under normal circumstances. When children were pointedly requested to explain their apparent fluctuations, they had to formulate their knowledge, and for this had to rely on and adapt their exiting concepts.
Summarizing, they had to perform two operations: (1) Extract the regularities from their experience: This part is akin to data mining techniques: factor analysis, cluster analysis, multi-dimensional scaling, where a number of dimensions, factors or clusters are identifiable (Purkhardt 1993), or connectionist systems, in which the weights on connections between units reflect the degree of statistical relatedness of the elements (Shanks 1997). (2) Formulate the regularity and interpreting it. This process involves already existing concepts and notions (Cheng 1997), that guide the process. Ultimately, they assimilated the process of sound traversal of materials to that of traversal of one material by another: it will take place when sound can seep through the obstacle, of if it can force its way through, and they did so by consciously formulating
In any event, since even adults do not construct extensive coherent accounts on most topics, the degree of coherence in children’s theories should not be exaggerated. Researchers and other practionners who engage in mapping cognitions find directed graphs or “conceptual structures” -- a graphic notation for typed first-order logic (Sowa 1992) -- very serviceable (Axelrod 1976; Laukkanen 1994; Laukkanen 1996; Sevón 1984) , to the point that a range of commercial software now exists to facilitate this type of mapping. The basic elements of such graphs are pairwise links between propositions or concepts. The “map” consists in bringing a set of such links together in a same graph. A common finding is that people typically refer only to pairwise links, or chains of such links. People can readily point at two quantities, and describe how changes in the one affect changes in the other. When these binary relations are combined into a map, patterns emerge, and these often come as a novelty to those very people who provided the binary relations. In that sense, the overall structure may not be ascribed to the respondents, who remain unaware of it. For instance, Williamson and Wearing (1996) mapped painstakingly the economic cognition of their subjects. they concluded that, whereas there exists an overall cognitive map that includes all the links mentioned by at least some subjects, individual subjects only mention small parts of that overall cognitive map, and none were aware of the whole. Some investigators have commented on how rare it is for subjects to mention features of the overall structure of their understanding, such as loops (Bougon & Weick 1977; Weick & Bougon 1986; White 1995a; White 1995b, Eden and Ackerman 1998) while Senge (1990) calls for “seeing interrelationships rather than linear cause-effect chains.”
The reason for this is that lay knowledge is pragmatic, goal-oriented: Given a proposition (a causal explanation, a generalization) - evaluate whether is it true. Given a situation, try and predict what will happen. Thinking is therefore narrowly focused, and satisfied with a local account. Studying the patterns formed by individually useful link is not useful. Cognitive maps are no longer considered as a representation of what is “in the mind” of the person who provided the individual links. On the contrary, drawing such maps may be used as a tool to help (usually) managers to get a grip on their own understanding, beneficial precisely because they don’t have an overall grasp of its structure (Eden ,1992 1994; Buzan 1993). This was already observed long ago by Evans-Pritchard (1937, cited in Moscovici 1992): common sense observes only certain links in the causal chain. Science observes them all, or at least most of them.
We argued that the requirement that coordination take place in working memory, coupled with the size limitation of working memory precludes large-scale integration. This is what must have amazed Marvin, the “paranoid android” of Adam’s fiction: the human cognitive system described here may seem so ineffectual as to be evolutionarily implausible, and this is a claim worth examining more closely. We commented on three deficiencies:
Explanatory systems are overly simple. As we saw, patterns of relative sizes of correlation may help us to formulate causal dependencies while conversely, a postulated mechanism helps us focus on the relevant relative correlations. Individual, local explanations are enough for solving local problems, and our mental apparatus handles them satisfactorily (White 1984). In Paleolithic times, humans were not expected to combine various explanatory schemata into a unified grand theory.
Decision making . Everyday decisions in pragmatic domains are usually handled very well by existing mechanisms. Remember Sperber’s distinction of intuitive and refbeliefs. Intuitive beliefs are derived, or derivable, from perception, and are limited to basic concepts referring to perceptually identifiable phenomena and to innately preformed unanalyzed abstract concepts. In such familiar domains, the help of decision support systems is not required … The failures of decision making we saw concern decisions in remote professional domains.
Coordination of ontologies of conceptual systems independently evolved is restricted. Since there is no known way of doing this, the failure of our cognitive system cannot constitute an argument. “Muddling through” may actually be the best policy. We might need to draw on what turn out to be a several disciplines to make a decision, several sets of considerations may be involved (as for instance, economics, where social, moral, economic, and interpersonal considerations all converge, Williamson and Wearing 1996; Watson 1996, or in the excerpt above of accounts of “why my child is epileptic”). When called upon to decide on a course of action, the best we can do may well be to select a dominant conception, or to weight several, without any attempt at deeper integration.
The limitations described here are real enough, and sometimes severe, but may have been evolutionary irrelevant.
Moore and Newell (1963) remarked: “A system understands knowledge to the extent it uses it whenever it is appropriate.” By that definition, there is little we really understand.
As we stated above, it is inappropriate to ask what is a subject’s understanding or naive theory on any given topic, because our knowledge relevant to a given topic does not form, psychologically, a natural kind.
How should research proceed? Another analogy may be useful here, this time for political sociology. Consider the question “what does Europe think about including Turkey in the European Union”, or “what does the Israeli people think about withdrawing from Lebanon?. These questions reflect legitimate concerns. Of course all the Europeans or the Israeli do not hold the same view, but a reasonable operationalization might produce a mean opinion, or a breakdown of various views. Much more appropriate however, would be to distinguish the various groups and currents that underlie the distribution of opinions. Thus, onecould see point at distinctions between respondents on the basis of their cultural background, economic circumstances, geography, religion, age and so forth. Such a finer analysis is not only important to generate a representative sample. More deeply, it is essential to understand the dynamics of the positions, and to predict their future evolution, on the basis of predictable demographic changes, economic and cultural development, educational intervention and so on.
The same state of affairs obtains intra-psychologically as well. It is not enough to obtain a respondent’s report on his or her naive theory. We must endeavor to identify the miscellaneous cognitive systems that contribute or constitute the naive theory, see when each is brought to bear, how they interact and what affects the outcome. The process of knowledge acquisition is such that lack of consistency is the normal state. If contradictions and inconsistencies are resolved in or under the control of Working Memory, a first central question is: What contradictions are tackled? This depends naturally on which ones are encountered, and which will be noticed, which in turn depends on what is retrieved from LTM: what retrieval cues are formulated and what do they retrieve, what types of associative pathways exist and how do they affect retrieval. But it also depends on how the subject manages the search and keeps track of the information retrieved.
The second central question is : How are inconsistencies resolved?. What are the mechanisms to resolve inconsistencies of various kinds, once they are noted
The subject who tries to build a coherent account may be likened to someone trying to put together the pieces of a puzzle into an image (Williams & Hollan 1981). Subjects may feel (FOK) that they have a fairly large set of pieces. But the image is misleading, and so may be their feeling. Not only are many pieces are missing, but the pieces belong to different puzzles, or to none, while the player is welcome to make new ones. Still, the metaphor captures an essential feature, one that Piaget called la raison constituante in contradistinction to la raison constituee: investigating a naive theory is studying a process more than it is studying a product.
Abelson, R. P., & Lalljee, M. (1988). Knowledge structures and causal explanation. In D. J. Hilton (Ed.), Seeing the connections in lay causal comprehension: A return to Heider (pp. 175-203). Brighton, UK: Harvester.
Aha, D. W. (1998). Lazy learning: Kluwer.
Anderson, J. R. (1983). The architecture of cognition. New York: Freeman.
Anderson, J. R. (1993). Rules of the Mind. Hillsdale, N.J.: Erlbaum.
Axelrod, R. (1976). The analysis of cognitive maps. In R. Axelrod (Ed.) Structure of Decision: The Cognitive Maps of Political Elites (pp. 55-72). Princeton, NJ: Princeton University Press.
Bachelard, G. (1940). La philosophie du Non. Paris: Presses Universitaires de France.
Baron, J. (1994). Thinking and deciding. (2nd ed.). Cambridge, UK: Cambridge U Press.
Baumard, Ph. (1996/1999) Tacit knowledge in organizations. London: Sage
Bazerman, M. H. (1994). Judgment in managerial decision making. (2nd ed.). New York: Wiley.
Bickhard, M. H., & Campbell, R. L. (1989). Interactivism and genetic epistemology. Archives de Psychologie, 57, 99-121.
Biswas, G., Schwartz, D., Bhuva, B., Bransford, J., Brophy, S., Balac, T., & Katzlberger, T. (1997). Analysis of student understanding of basic electricity concepts (Technical Report TR-CS-97-01). Nashville, TN: Vanderbilt University
Bloch, M. (1998). How we Think they Think, Anthropological Approaches to Cognition, Memory and Literacy: West-View.
Bougon, & Weick, K. (1977). Cognition in Organizations. Administrative Science Quartely, 606-639.
Bourdieu, P. (1990). The logic of practice. Stanford, Calif.: Stanford University Press.
Breakwell, G. M., & Canter, D. V. (Eds.). (1993). Empirical approaches to social representations. Oxford, England: Clarendon Press/Oxford University Press.
Brown, A. L. (1989). Analogical learning and transfer: What develops? In S. Vosniadou & A. Ortony (Eds.), Similarity and analogical reasoning (pp. 369-412). Cambridge, MA: Cambridge University Press.
Buzan, T. (1993) The Mind Map Book. London: BBC Books
Carey, S. (1991). Knowledge acquisition: Enrichment or conceptual change? In R.Gelman & S.Carey (Eds.), the epigenesis of mind: Essays on biology and cognition. Hillsdale, NJ.: Erlbaum.
Carey, S., & Gelman, R. (Eds.). (1991). Epigenesis of mind: Essays in biology and knowledge. Hillsdale N.J.: Erlbaum.
Carey, S., & Spelke, E. (1994). Domain-specific knowledge and conceptual change. In L. A. Hirschfeld & S. A. Gelman (Eds.), Mapping the Mind: domain specificity in cognition and culture (Vol. 7). Cambridge: Cambridge U. Press.
Cheng, P. W. (1997). From covariation to causation: a causal power theory. Psychological Review, 104: 367-405.
Chi, M. T. H. (1992). Conceptual change within and across ontological categories: examples from learning and discovery in science. In R. N. Giere (Ed.), Cognitive models of science (Vol. 15, pp. 129-186). Minneapolis: University of Minnesota Press.
Clark, A., & Toribio, J. (1998). Cognitive architectures in artificial intelligence: the evolution of research programs. New York: Garland.
Cooke, N.J. and S.D. Breedin. (1994). Constructing naive theories of motion on the fly. Memory and Cognition 22: 474-493.
D’Andrade, R. (1995) The development of cognitive anthropology.Cambridge, UK: Cambridge University Press.
D'Andrade, R. (1995). The development of Cognitive Anthropology: Cambridge University Press.
Dienes, Z., & Perner, J. (1999). A Theory of Implicit and Explicit Knowledge. Behavioral and Brain Sciences, 22(5).
DiSessa, A. (1997). Speculations on the foundations of knowledge and intelligence. In D. Tirosh (Ed.), Implicit and explicit knowledge: and educational appro. Norwood, NJ: Ablex.
diSessa, A., & Sherin, B., L. (in press). What changes in conceptual change. International Journal of Science Education.
Doise, W. (1989). Cognitions et reprentations sociales: l’approche genetique. In D. Jodelet (Ed.), Les representations sociales. Paris: PUF.
Duveen, G., & Lloyd, B. (Eds.). (1990). Social representations and the development of knowledge. Cambridge, England: Cambridge University Press.
Eden, C. (1992). On The Nature of Cognitive Maps. Journal of Management Studies, 29: 261-265.
Eden, C. (1994). Cognitive mapping and problem structuring for system dynamics model building. Special Issue: Systems thinkers, systems thinking. System Dynamics Review, 10: 257-276.
Eden, C. and Ackerman, F. (1998). Making Strategy. London: Sage
Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1996). Rethinking innateness: a connectionist perspective on development. Cambridge, MA: MIT Press.
Evans-Pritchard, B.E. (1937) Witchcraft, Oracles and magic among the Azande. Oxford: Oxford University Press
Fodor, J. A. (1983). The modularity of mind. Cambridge, M.A.: MIT.
Foucault, M. (1972). The archaeology of knowledge.. New York,: Pantheon Books.
Furth, H. G. (1975). Piaget and Knowledge. New York: Prentice Hall.
Gardner, H. (1983). Frames of mind: the theory of multiple intelligences. New York: Basic.
Genter, D., & Stevens A.L. (Eds.) (1983). Mental Models. London , Lawrence Erlbaum Associates
Gentner, D., & Toupin, C. (1986). Systematicity and surface similarity in the development of analogy. Cognitive Science, 10: 277-300.
Gentner, D., Rattermann, M. J., & Forbus, K. D. (1993). The roles of similarity in transfer: Separating retrievability from inferential soundness. Cognitive Psychology, 25: 524-575.
Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive Psychology, 12: 306-355.
Gigerenzer, G. (1996). From tools to theories: Discovery in cognitive psychology. In C. F. Graumann, K. J. Gergen, & e. al. (Eds.), Historical dimensions of psychological discourse. (pp. 36-59). New York, NY: Cambridge University Press.
Gigerenzer, G., & Goldstein, D. G. (1996). Mind as computer: Birth of a metaphor. Creativity Research Journal, 9.
Gigerenzer, G., & Regier, T. (1996). How do we tell an association fra rule? Comment on Sloman (1996). Psychological Bulletin, 119(1): 23-26.
Glymour, C. (1998). Learning causes: Psychological explanations of causal explanations. Minds and Machines, 8(1): 39-60.
Godden, D., & Baddeley, A. D. (1975). Context-dependent memory in two natural environments: on land and under water. British Journal of Psychology, 66: 325-331.
Goodwin, P., & Wright, G. (1991). Decision analysis for management judgment. New York: Wiley.
Gopnik, A., & Wellman, H. M. (1992). Why the child’s theory of mind really is a theory. Special Issue: Mental simulation: Philosophical and psychological essays. Mind and Language, 7: 145-171.
Gruber, T. (1995). Towards principles for the design of ontologies used for knowledge sharing. International Journal of Human--Computer Studies, 43(5/6): 907-928.
Gruber, T. R. (1993). A translation approach to portable ontologies. Knowledge Acquisition, 5(2): 199-220.
Guarino, N., & Giaretta, P. (1995). Ontologies and knowledge bases- towards a terminological clarification. In N. J. Mars (Ed.), Towards Very Large Knowledge Bases - Knowledge Building and Knowledge Sharing (pp. 25-32). Amsterdam: IOS Press.
Helman, C. G. (1978). Feed a cold, starve a fever - Folk models of infection in an Enlglish suburban community, and their relation to medical treatment. Culture, Medicine and Psychiatry, 42: 107-137.
Higgins, E. T. (1996). Knowledge activation: accessibility, applicability, and salience. In E. T. Higgins and A. Kruglanski (Eds.) Social psychology: Handbook of basic principles. (133-168) New York: Guilford Press.
Hintzman, D. L. (1986). "Schema abstraction" in a multiple-trace memory model..
Hirschfeld, L. A. & S. A. Gelman (1994) (Eds.), Mapping the Mind: domain specificity in cognition and culture. Cambridge: Cambridge U. Press.
Holyoak, K. J., & Koh, K. (1987). Surface and structural similarity in analogical transfer. Memory and Cognition, 15(4): 332-340.
Holyoak, K. J., & Simon, D. (1999). Bidirectional reasoning in decision making by constraint satisfaction. Journal of Experimental Psychology : General, 128(1): 3-31.
Holyoak, K. J., Novick, L. R., & Meltz, E. (1994) Component processes in analogical transfer: mapping, pattern completion and adaptation. In K. J. Holyoak & J. A. Barnden (Eds.), Advances in connectionist and neural computation theory. (Vol. 2: analogical connections, pp. 113-180). Norwood, NJ: Ablex.
Johnson, S. C., & Carey, S. (1998). Knowledge enrichment and conceptual change in folkbiology: Evidence from Williams syndrome. Cognitive Psychology, 37(2) 156-200.
Kaiser, M.K., D.R. Proffitt, S. M. Whelan and H. Hecht. (1992). Influence of animation on dynamical judgments. Journal of Experimental Psychology: Human Perception and Performance, 18: 384-393.
Karmiloff-Smith, A. (1992). Nature, Nurture and PDP: Preposterous Developmental Postulates? Connection Science, 4(3): 253-269.
Karmiloff-Smith, A. (1993). Beyond modularity: a developmental perspective on cognitive science. Cambridge, MA: MIT Press.
Karmiloff-Smith, A., & Clark, A. (1993). What’s special about the development of the human mind/brain? Mind and Language, 8(4): 569-581.
Keil, F. C. (1990). Constraints on constraints: Surveying the epigenetic landscape. Cognitive Science, 14(1): 135-168.
Keil, F. C. (1994). The birth and nurturance of concepts by domains: the origins of concepts of living things. In L. A. Hirschfeld & S. A. Gelman (Eds.), Mapping the Mind: Domain specificity in cognition and culture. Cambridge: Cambridge U. Press.
Knowlton, B. (1997). Declarative and nondeclarative knowledge: insights from neuroscience. In K. Lamberts & D. R. Shanks (Eds.), Knowledge, concepts and categories (pp. 215-246). Hove, UK: Psychology Press.
Koriat, A. (1994). Memory’s knowledge of its own knowledge: The accessibility account of the feeling of knowing. In J. Metcalfe & A. P. Shimamura (Eds.), Metacognition: Knowing about knowing. Cambridge, MA: The MIT Press.
Koriat, A. (1995). Dissociating knowing and the feeling of knowing: Further evidence for the accessibility model. Journal of Experimental Psychology General, 124(3): 311-333.
Kruglanski, A. W. (1989). Lay epistemics and human knowledge : cognitive and motivational bases. New York: Plenum Press.
Kunda, Z., & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A Parallel-constraint- satisfaction theory. Psychological Review, 103(2): 284-308.
Laukkanen, M. (1994). Comparative cause mapping of organizational cognitions. Special Issue: Cognition within and between organizations: Five key questions. Organization Science, 5: 322-343.
Laukkanen, M. (1996). Conducting Cause Mapping Research: Opportunities and challenges. In B. Wintersheid & C. Eden (Eds.), Managerial and Organizational Cognition: new directions in theory, methods and research.. London: Sage.
Leiser, D. (1996). Computational emergence and development emergence: a Sobering survey. Intellectica, 22: 203-227.
Leiser, D. (1997). The dynamics of cumulative knowledge. Behavioral and Brain Sciences, 20(1), 76.
Leiser, D., Doitsch, E., & Meyer, J. (1996). Mothers' lay models of the causes and treatment of fever. Social Science & Medicine, 43(3): 379-387.
Leiser, D., Sevon, G., & Levy, D. (1990). Children's economic socialization: Summarizing the cross-cultural comparsion of ten countries. Special Issue: Economic socialization. Journal of Economic Psychology, 11(4): 591-631.
Logan, G. D. (1988). Toward an instance theory of automatization. Psychological Review, 95(4): 492-527.
Markman, A. B., & Dietrich, E. (1998). In Defense of Representation as Mediation. Psycoloquy, 9(48).
Marshall, J. C. (1977). Minds, machines and metaphors. Social studies of science, 7: 475-488.
Mazens, K., & Lautrey, J. (1999). Etude exploratoire des representations "naives" des sons chez les enfants de 6 a 10 ans. Revue de Psychologie de l'Education.Xx-xx
McCloskey, M. (1983). Intuitive physics. Scientific American 248: 122-130.
McGuire, W. J. (1990). Dynamic operations of thought systems. American Psychologist, 45: 504-512.
Metcalfe, J., & Shimamura, A. P. (Eds.). (1994). Metacognition: Knowing about knowing. Cambridge, MA: The MIT Press.
Moore, J., & Newell, A. (1963). How can MERLIN understand? In L. W. Gregg (Ed.), Knowledge and Cognition (pp. 201-252): Hillsdale: Erlbaum.
Morton, J., Hammersley, R. H., & Bekerian, D. A. (1985). Headed records: A model for memory and its failures. Cognition, 20(1): 1-23.
Moscovici, S. (1990). Extending the conversation. In G. Duveen & B. Lloyd (Eds.), Social Representations and The Development Of Knowledge. Cambridge: Cambridge U Press.
Moscovici, S. (1992) La mentalite prelogique des civilises. in U. Flick (Ed.) La perception quotidienne de la sante et de la maladie – Theories subjectives et representations sociales. Paris: l’Harmattan.
Mulhall, A. (1996). Cultural discourse and the myth of stress in nursing and medicine. International Journal of Nusing Studies, 33(5): 455-468.
Newell, A. (1990). Unified theories of cognition. Cambridge, Mass.: Harvard University Press.
Nussbaum, J., & Sharoni-Dagan, N. (1983). Changes in second grade children's preconceptions about the earth as a cosmic body resulting from a short series of audio-tutorial lessons. Science education, 67: 99-114.
Pennington, N., & Hastie, R. (1988). Explanation-based decision making: effects of memory structure on judgment. Journal of Experimental Psychology Learning, Memory and Cognition, 14(3): 521-533.
Piaget, J. (1945/51) “Play, dreams and imitation in childhood” (1951; New York: Norton)
Piaget, J. (1974) Adaptation vitale et psychologie de l'intelligence. Hermann.
Piaget, J. (1974/1978). Success and understanding. London: Routledge & Kegan Paul.
Piaget, J. (1977/1985). The equilibration of cognitive structures : the central problem of intellectual development. Chicago: University of Chicago Press.
Piaget, J., & Garcia, R. (1971/1974). Understanding causality. New York: Norton.
Purkhardt, S. C., & Stockdale, J. E. (1993). Multidimensional scaling as a technique for the exploration and description of a social representation. In G. M. Breakwell & D. V. Canter (Eds.), Empirical approaches to social representations (p. 272-297). Oxford, England: Clarendon Press/Oxford University Press.
Ranney, M. (1994). Relative consistency and subjects' "theories" in domains such as naive physics: common research difficulties illustrated by Cooke and Bredin. Memory and Cognition, 22(4): 494-502.
Ranney, M. (1995). Explorations in explanatory coherence. In E. B. On, B. Eylon, & Z. Schertz (Eds.), Designing intelligent learning environment.. Norwook, NJ: Ablex.
Ranney, M., & Schank, P. (1998). Toward an integration of the social and the scientific: Observing, modeling, and promoting the explanatory coherence of reasoning. In S. J. Read, L. C. Miller et al. (Eds.), Connectionist models of social reasoning and social behavior (pp. 245-274). Mahwah, NJ, USA: Lawrence Erlbaum.
Read, S. J., & Miller, L. C. (1993). Rapist or regular guy: explanatory coherence in the construction of mental models of others. Personality and Social Psychology Bulletin 19: 526-541.
Read, S. J., & Miller, L. C. (1994). Dissonance and balance in belief systems: the promise of parallel constraint satisfaction processes and connectionist modeling approaches. In R. C. Schank & E. Langer (Eds.), Beliefs, reasoning and decision making (pp. 209-237). Hillsdale, N.J.: Lawrence Erlbaum Associates.
Reder, L. M., & Ritter, F. E. (1992). What determines initial feeling of knowing? Familiarity with question terms, not with the answer. Journal of Experimental Psychology: Learning Memory and Cognition, 18: 435-451.
Rosser, R. (1994). Cognitive development - Psychological and biological perspectives. Boston: Allyn and Bacon.
Ryle, G. (1949). The concept of mind. London: Norton.
Schwarz, N. (1995). Social Cognition: information accessibility and use in social judgment. In E. E. Smith & D. N. Osherson (Eds.), Thinking: an invitation to cognitive science (Vol. 3, ). Cambridge, MA: MIT.
Schwarz, N. (1999). Self-reports - How the questions shape the answers. American Psychologist, 54(2), 93-105.
Senge, P. M. (1990). The fifth discipline : the art and practice of the learning organization. New York: Doubleday
Sevón, G. (1984). Cognitive maps of past and future economic events. Acta Psychologica, 56: 71-79.
Shanks, D. (1997). Distributed representations and implicit knowledge. In K. Lamberts & D. R. Shanks (Eds.), Knowledge, concepts and categories (pp. 215-246). Hove, UK: Psychology Press.
Shanon, B. (1976). Aristotelianism, Newtonianism, and the physics of the layman. Perception 5: 241-243
Sher, R. (1999). Naive theories on couple formation and maintenance. Unpublished MA thesis, Ben-Gurion University, Beer Sheva.(Hebrew)
Shimamura, A. P. (1986). Priming effects in amnesia: Evidence for a dissociable memory function. Quarterly Journal of Experimental Psychology, 38A: 619-644.
Shultz, T. R., & Lepper, M. R. (1992). A constraint satisfaction model of cognitive dissonance phenomena. Paper presented at the 14th Annual Conference of the Cognitive Science Society.
Sloman, S. A. (1996a). The empirical case for two systems of reasoning. Psychological Bulletin, 119(1): 3-22.
Sloman, S. A. (1996b). The probative value of simultaneous contradictory belief: Reply to Gigerenzer and Regier (1996). Psychological Bulletin, 119(1): 27-30.
Sowa, J. F. (1992). Conceptual analysis as a basis for knowledge acquisition. In R. R. Hoffman (Ed.), The Cognition of Experts: Psychological Research and Empirical AI. Berlin.: Springer.
Sparrow, J. (1998) Knowledge in 0rganization: access to thinking at work. London: Sage
Spelke, E. S., Breinlinger, K., Macomber, J., & Jacobson, K. (1992). Origins of knowledge. Psychological Review, 99(4): 605-632.
Sperber, D. (1990). The epidemiology of beliefs. In C. Fraser & G. Gaskell (Eds.), The social psychology of widespread beliefs. Oxford: Clarendon Press.
Sperber, D. (1997). Intuitive and reflective beliefs. Mind & Language, 12: 67-83.
Squire, L. R., Knowlton, B. J., & Musen, G. (1993). The Structure and Organization of Memory. Annual Review of Psychology, 44: 453-495.
Stainton-Rogers, W. (1991). Explaining health and illness: An exploration of diversity. London: Harvester-Wheatsheaf.
Strauss, C., & Quinn, N. (1998). A Cognitive Theory of Cultural Meaning: Cambridge University Press.
Thagard, P. (1989). Explanatory Coherence. Behavioral and Brain Sciences, 12: 435-502.
Thagard, P. (1992). Conceptual Revolutions. Princeton, New Jersey: Princeton University Press.
Thagard, P. (1998). Explaining disease: correlations, causes and mechanisms. Minds and Machines, 8: 61-78.
Toulmin, S. (1958). The Uses of Argument: Cambridge, UK, Cambridge University Press.
Tulving, E., & Osler, S. (1968). Effectiveness of retrieval cues in memory for words. Journal of Experimental and Psychology, 77: 593-601.
Uschold, M., & King, M. (1996). Towards a Methodology for Building Ontologies. Knowledge Engineering Review, 11(2).
Vermersch, P. (1979). Peut-on utiliser les donne0es de la psychologie genetique pour analyser le fonctionnement cognitif des adultes - Theorie operatoire de l'intelligence et registres de fonctionnements. Cahiers de Psychologie, 22: 59-74.
Vigotsky, L. S. (1934/1986). Thought and Language (Alex Kozulin, Trans.). 2nd (revised ). Cambridge, MA: The MIT Press.
Vosniadou, S., & Brewer, W. F. (1992). Mental models of the earth: A study of conceptual change in childhood. Cognitive Psychology, 24(4): 535-585.
Watson, T. J. (1996). How do managers think? Identity, morality and pragmatism in managerial theory and practice. Management Learning, 27(3): 323-341.
Weick, K. E., & Bougon, M. G. (1986). Organisations as Cognitive Maps: Charting Ways to Success and Failure. In H. P. Sims & D. A. Gioia (Eds.), The Thinking Organisation. San Francisco: Jossey-Bass.
Wellman, H. M., & Gelman, S. A. (1992). Cognitive development: Foundational theories of core domains. Annual Review OF Psychology, 43: 337-375
Wellman, H. M., & Inagaki, K. (Eds.). (1997). The emergence of core domains of thought: Children's reasoning about physical, psychological, and biological phenomena. New directions for child development. San Francisco, CA, USA: Jossey-Bass Inc, Publishers.
White, P. (1984). A model of the layperson as pragmatist. Personality and Social Psychology Bulletin, 10: 333-348.
White, P. A. (1995a). Common-sense construction of causal processes in nature: A causal network analysis. British Journal of Psychology, 86(3): 377-395.
White, P. A. (1995b). Use of prior beliefs in the assignment of causal roles: Causal powers versus regularity-based accounts. Memory and Cognition, 23(2): 243-254.
Wilkes, A. L. (1997). Knowledge in minds. Hove, UK: Psychology Press.
Williams, M. D., & Hollan, J. D. (1981). The process of retrieval from very long-term memory. Cognitive Science, 5: 87-119.
Williamson, M. R., & Wearing, A. J. (1996). Lay people’s cognitive models of the economy. Journal of Economic Psychology, 17(1): 3-38.
Wilson T D and Hodges S D 1992 Attitudes as temporary constructions. In L.L. Martin & A. Tesser (Eds.) The construction of social judgments (37-65) Hillsdale, NJ: Erlbaum.
Yates, J., Bessman, M., Dunne, M., Jertson, D. & et al. (1988). Are conceptions of motion based on a naive theory or on prototypes? Cognition, 29: 251-275
Zani, B. (1993). Social representations of mental illness: Lay and professional perspectives : In Breakwell, G. M., & Canter, D. V. (Eds.). Empirical approaches to social representations. Oxford, England: Clarendon Press/Oxford University Press.
Endnotes
Email: dleiser@bgumail.bgu.ac.il. Address: Dept of Behavioral Sciences, Ben-Gurion University, PO Box 653, Beer Sheva 84105, Israel. The contributions of Tamar Lavi, Ronit Sher, Yossi Mor-Yossef, Shelly Drori, Ayelet Solal, Ayelet Cassuto, Daphna Eisen and Liat Beck are gratefully acknowledged.
2 Actually, this is not the case. Roy d’Andrade reports the results of scientific studies that fail to demonstrate such an effect, even though many Americans believe it is real, but he does not explain the origin of this misconception. The need to reconcile incompatible beliefs readily accounts for it.
3 His book called in English “Play, dreams and imitation in childhood” is called in French “La formation du symbôle chez l'enfant: imitation, jeu et rêv, image et représentation” that is: “The formation of the symbol in the child: Imitation, play and dream, image and representation”. The badly maimed title was presumably thought to be more saleable.
4 Karmiloff-Smith too acknowledges that when knowledge is directly encoded in linguistic form, the process is different.
5 These might be identified with the outcome of reflective abstraction, as discussed in the previous section.
6 See e.g. http://www.tzi.de/grp/i3/projects-class_fr.html; http://www.irit.fr/GRACQ/
7 See http://rombutan.mit.edu/arpa.html
8 Ontologies in developing knowledge is a profound domain (see e.g., Foucault 1972 and diSessa, & Sherin in press for a recent psychologically informed effort). For our purposes, we may rely on the simpler way the term is used in Artificial Intelligence, in the contexts of knowledge acquisition and of knowledge sharing: an ontology is a specification of a conceptualization, a description of the concepts and relationships that can exist for an agent (Gruber 1993). A more precise analyses distinguishes two meanings : (1) a conceptualisation: an intensional semantic structure which encodes the implicit rules constraining the structure of a piece of reality; and (2) an ontological theory: a set of formulas intended to be always true according to a certain conceptualisation (Guarino and Giaretta,1995).
9 We are concerned in this section with general mechanisms for the harmonization of knowledge. There exist also other specialized mechanisms. For instance, Kruglanski (1989) showed that various effects (the algebra of love/hate relations, aggression/support, halo effects, dissonance etc) are all fundamentally issues of logical coherence. On the other hand, it is perfectly possible that each of these special cases is represented by a different neural, even hormonal system. The coherence mechanisms could be specialized and distinct, despite their having a same logical or even computational structure.