Coursehero >> Canada >> Université du Québec à Montréal >> DIC 8101

Course Hero has millions of student submitted documents similar to the one below including study guides, homework solutions, papers, exam answer keys and textbook solutions.

De 2 Domain and the Basic QuestiO18 The domain-specifying assumptionsand the basic questions of a research framework tell us what scientistscommitted to that framework are interested in understanding and explaining at least at the outset of their inquiries , . Cognitive scientistsare not entirely in agreementas to the nature of the domain and the basic questionsfor cognitive science(or , to be more , precise of the subsidiary research framework of cognitive science that dealswith adult , normal , typical cognition). In this chapter, I will chart and attempt to adjudicate the various disagreements and will present what I , take to be the most defensibleconception. The domain-specifying assumptions of a research framework include three kinds of claims: an identification assumption which indicates what , kinds of phenomena are to be included in the domain; property assumptions , which specify essential or particularly important properties of the phenomena identified by the identification assumption and a grouping ; assumption, which indicates that the phenomenapicked out by the identification assumption make up a theoretically coherent set of phenomena , or a system Each of these three kinds of claims will be discussed in . turn . Section 2 1 will take up the identification assumption section 2 will . .2 , discuss various property assumptions and section 2.3 will discuss the , will concern the precise grouping assumption. Most of the disagreements nature of the identification assumption for ANTCOG . However, one further area of controversy will emerge this will concern the best way to ; formulate the grouping assumption. 2 TheIdentification .1 Assumption Everyone agreesthat the domain of cognitive scienceis human cognition or intelligence However, there is controversy regarding precisely what . of cognition (or intelligence are or should be the focusesof study. aspects ) On my view, the best way to state the identification assumption of cognitive scienceis this: D 1 (IDENTIFICA ASSUMPTION nON ) The domain of ANTCOG consists of the human cognitive capacities . Although it is not very clear in the pretheoretic schemeof things precisely what a cognitive capacity is (and, in particular , what distinguishes cognitive from noncognitive capacities there does seemto befairlywide ), Chapter 2 spread agreement on what constitutes the clear caseswithin the general . class They include our capacity to uselanguage(perceiveit , comprehend it, produce it, translate it , communicate with it, etc.), to perceivevisually, to apprehend music, to learn, to solve problems (reason draw inferences, , ) to plan action, to act intentionally , to remember and to imagine. Each of , these subclass can of course be broken down into far more specific es , , ; , capacities for example the capacity to rememberincludes the capacity to remember faces episodesfrom the past, lists of nonsensesyllables facts, , , and so on. There are also severalborderline phenomenawhose , concepts , membership in the domain of cognitive science is, at present unclear. Theseinclude our capacity to acquire skills with a significant motor component and our capacity for nonlinguistic auditory , tactile, olfactory, and gustatory perception. I said above that there was controversy over Dl . This does not mean , however that I think the evidential basisfor is Dl weak. To the contrary, I , believe that if one looks at what cognitive scientists actually study- the subject matter of their research it becomesevident that every clear case of cognitive scienceresearchfocuseson one or another of the cognitive . capacities To provide a completely persuasivedescriptive argument in support of this contention would require counting and categorizing every pieceof cognitive scienceresearchthat has beendone to date. For obvious reasons I do not propose to do that here. However, short of such an , exhaustivesurvey, a compelling inductive casecan be made by inspection of any of the currently available surveysof the field. Consider, for example , Gardner' s 1985book The Mind 's New Sciencein part II of which Gardner , discuss several researchefforts that , on his view, " qualify for the label es cognitive scientific, . . . involve the central questionsin cognitive science[ ,] and . . . are consideredby the cognitive-scientific community to be of high " ' quality (pp. 293- 294). Theseinclude David Marr s work on human visual ' s research on visual , perception Stephen Kosslyn imagery, the work of Eleanor Rosch Brent Berlin, and Paul Kay on color naming and categorization , , and the researchof Amos Tversky, Daniel Kahneman, and Philip Johnson- Laird on human reasoning Note that each of these research . efforts focuseson one or another of the cognitive capacities specifically , the capacity to perceivevisually, to usevisual images to categorizecolors, , and to reason . Despite the fact that the nature of cognitive scienceresearch clearly supports Dl , one does occasionallyfind somewhatdifferent (though close The Domain and the Basic Questions ly related) characterizations of the domain of cognitive science in the ' literature. That is, cognitive scientists metascientific claims sometimes diverge from 01 . There are four principal alternatives: is Alternative 1 The domain of cognitive science human intelligent behavior or action. Alternative 2 attitudes . science thehumanpropositional is Thedomainof cognitive Alternative 3 The domain of cognitive science is human knowledge representation and use. Alternative 4 The domain of cognitive science consists only of those human cognitive capacities that are unencapsulated- that is, those whose " ' " exercise is sensitive to the subject s goals and general knowledge . None of these alternatives is radically off - base. However , let me indicate why I think that noneof themreallyfits the bill. Alternative 1 The view that the domain of cognitive scienceis human intelligent behavior or action tends to go hand in hand with the position, adopted by many , people in artificial intelligence that the domain of cognitive scienceis . in general including the intelligence of nonhuman machines1 , intelligence I shall not be dealing with the question As I noted in the introduction, 's . of extending cognitive science domain to nonhuman subjects I would, however, like to consider whether, given that we restrict the domain to the is human speciesthe aim of cognitive science bestconceivedof as the study , of human intelligent behavior. Oearly , from the point of view of someonewho endorsesDI , there is somethingright about alternative I . The exerciseof any given capacity can be viewed as a piece of intelligent behavior or action (albeit sometimesa mental one), and DI claims that the domain of cognitive scienceis the . human cognitive capacities For example the exercise of my capacity , to read constitutes an act of reading and the exerciseof my capacity to , answer questions about a text constitutes an act of question-answering ; and both of theseare surely instancesof intelligent behavior. Nevertheless there are severalthings that trouble me about conceiving , of the domain in this way. First, I have been unable to discover a concep - Chapter 2 tion (or a theory) of intelligence that doesnot subject alternative 1 to fairly . obvious counterexamples Second alternative 1 naturally gives rise to several , . unfortunate methodological tendencies And third , it raises false expectati ' about the aims of cognitive sciences researchprogram. What, then, is intelligent behavior supposedto be? If one examineswhat has been said on the subject, one immediately notices that there are two quite different conceptionsof intelligent behavior at work . Intelligence is a comparative notion, and it appears that the comparison can either be . interspecies or intraspecies When people within the field of cognitive science (such as Simon [ 1981ab] or Haugeland [ 1985 ) write about , ] , they seemto be thinking about the various properties studying intelligence of mind (and the behaviors to which they give rise) that distinguish . human beings(and certain nonhuman machines from lower species Pre) , sumably this is also the notion of intelligent behavior that Pylyshyn ( 1984 . 3) has in mind when he gives the following as an example of the sort of p " " phenomena cognitive scienceseeksto explain: . turnsandstarts A pedestrian walkingalonga sidewalkSuddenly pedestrian is the to crossthe street At the sametime a car is travelingrapidly down the street . , . the . The driver of the car applies brakesThe car skidsand towardthe pedestrian . hesitates swerves over to the side of the road hitting a pole The pedestrian , , ' . thengoes overandlooksinsidethecar on thedrivers side He runsto a telephone 9 boothat thecomeranddialsthe numbers and 1. In contrast, when psychologists (such as Sternberg [ 1984 ) speak about ] have in mind studying only those properties of mind (and intelligence, they the behaviors to which they give rise) that make a person " smart." Let me call the former the " democratic" conception of intelligent behavior and the latter the " elitist " conception. Any version of alternative 1 formulated in terms of an elitist conception of intelligent behavior will clearly result in too narrow a view of the domain of cognitive science This is becausecognitive scientistsare interested . in studying cognitive capacities such as the capacity to perceiveor - that all normal human beings have in perform an intentional action of common, no matter how intelligent (in the sense smart) or unintelligent in the senseof stupid) they might be.2 Unfortunately , the concept of ( intelligent behavior in the democratic sense(in which all normal human beings are intelligent) has received very little attention in the literature. I have come across only two treatments (there may, of course be others), , neither of which is very satisfactory. The Domain and the Basic Questions In his book Artificial Intelligence: The Very Idea, Haugeland ( 1985 , ' " . 6) raisesthe question of what intelligence is only to dismiss it : Doesn t p , everything turn on this? Surprisingly, perhaps very little . . . . For practical , ) purposes a criterion proposed by Alan Turing ( 1950 satisfiesnearly everyone .,,3 The final remark is a perplexing one, for the Turing test was proposed as a criterion for when a machinethinks. It does not , in and of itself, to the question of what the explanandum of cognitive scienceshould speak be, and Haugeland does not tell us how to adapt it for this purpose. What Haugeland may have in mind is something like this: The aim of cognitive scienceis to explain the sort of intelligent verbal behavior in humans that , if generatedby a machine, would allow it to passthe Turing test. But surely this won' t do. In the first place, not everyonebelievesthat passing the Turing test ought to be taken as a sufficient condition for a machine' s being intelligent (see for example Block 1981band Bieri 1988. , , ) Second even if one believesthat the Turing test constitutes an adequate , ' criterion for a systems being intelligent, what is criterial is that the system " " a generates body of behavior sufficiently rich to induce a judge to believe that the systemis a human being. However, no individual pieceof behavior need be part of this larger body of behavior. Hence the adapted criterion , does not really speak to the question of when an individual piece of behavior is intelligent . Finally , the suggestion (in adapted form ) would restrict us to verbal behavior, whereaswe surely want to include nonverbal . intelligent behavior in the domain of cognitive science The question of what intelligent behavior is fares somewhat better in Simon's hands, but there are still difficulties. Simon ( 1981a p. 15 writes: , ) their exhibit their intelligence achieving goals(e.g., meeting by Intelligentsystems environments . Intelligent needs survival in the faceof differentand changing for ) behavioris adaptive hencemust take on strikinglydifferentforms when the , . environments correspondingly are different ' Severalpoints in this passage require clarification . First , Simon uses envi' to mean either an external or an internal environment. Second ronment , the referenceto " goals" might appear to be mentalistic, Simon although ' seemsto have nothing more in mind than the notion of a systems " function " - the function that a devicecan be designedto have or that a biological organism can acquire by evolution. I infer this on the basisof a passage b on page 15 of Simon 1981 in which goals are attributed to a relatively machine (a so-called motor controller ). Finally , Simon (1981a pp. , simple Chapter 2 15- 16) takes it that adaptation can occur on three time scales: On the shortest time scale intelligent - hence adaptive- systemscontinually , changetheir behavior in the courseof solving each problem situation they encounter .. . . On a somewhatlonger time scale intelligent systemsmake adaptations that are , fully preservedand remain available for meeting new situations success . They learn. There are many forms that this semi-permanent adaptation can take, and corresponding many forms of learning. One important form is the accumulation of routes for retrieving it . . . . information in memoriesand the acquisition of access On the longest time scale intelligent systemsevolve. Their evolution may be , Darwinian .. . . It may equally well be social, through the discovery of new knowledge and strategiesand their transmission from one systemto another. There are two questions that might be raised about Simon' s account: Is his conception of intelligent behavior intuitively plausible- that is, does it properly delineate between intuitively clear casesof intelligent behavior and intuitively clear casesof nonintelligent behavior? Does it correspond to the class of phenomena cognitive sciencestudies or ought to study? , Since I doubt that we have an intuitively clear idea of what intelligent . behavior is, I will consider only the secondof thesequestions 's Alternative 1, formulated in terms of Simon conception of intelligent behavior, seemstoo broad, for there are casesof adaptive behavior (in Simon' s sensemediated by fairly low -level biological mechanismsthat are ) not in the least bit cognitive and, hence that do not belong within the , domain of cognitive science An interesting example is the phenomenon . called " conditioned taste aversion." It has been experimentally demonstrated in rats (Garcia Ervin, and Koelling 1966 and in children (Bernstein , ) 1978 that if a subject eats something and becomesill shortly thereafter he , ) or shewill tend to avoid that food in the future. If we consider the goal of the organism in this exampleto be somethinglike to ingest food conducive to health and survival, the casefits Simon' s conception quite nicely. The goal is to ingest food conducive to survival; the changing environment consistsof changing food options: the behavior consistsin responding to one of those food options (an apparently unhealthy one) by rejecting it . That rejection is adaptive with respect to the goal in question. But although this fits Simon' s conception of intelligent behavior, it is clearly as outside the domain of cognitive science the latter is currently construed. Moreover, there is reason to believe that this is how matters should be: In the experimentswith the children evenwhen the children knew perfectly , The Domain and the Basic Questions well that the nausea they were experiencing was not caused by the ice cream but was only associatedwith it temporally, they still rejectedthe ice cream. Their cognitive apparatus could not overrule the biological mechanism , responsiblefor the avoidance response(Mook 1987 p. 239). The idea of intelligent behavior rests on two notions: the notion of intelligenceand the notion of behavior. Thus far, my objections to alternative 1 have turned on the fact that there is currently (to my knowledge ) no conception of intelligencethat makesthe claim a plausible one. I should now like to point out that, even if an adequateconception of intelligence were to be found, alternative 1 would still not be optimal , becauseit restricts the domain of cognitive scienceto certain forms of behavior . An obvious difficulty is that alternative 1 seemsto rule out the exercise of all capacitiesthat are either purely input capacities(such as perception) or purely " inner" capacities (such as memory). There is, however, a way around this difficulty , although it is not very satisfactory. Alternative 1 restricts us to the exerciseof capacitieswith an output component only at the level of the explanandum. Phenomena such as perception and . memory could still be part of the picture at the level of the explanans For , example although we are barred from regarding perception as part of the domain, we are free to include perceptual discriminative behavior. And in giving our explanations of discriminative behavior, we are free to invoke our perceptualcapacities Although this move probably works to preserve . most of what we want within the " grasp" (if not , technically, within the domain) of cognitive science it points up two unfortunate methodological , featuresof conceiving of the domain in this way. The first is that a research framework dedicated to explaining behavior rather than the cognitive , capacitiesmay have sometendencyto overlook (and, hence never account for) one of the central properties of those capacities namely, their intenbehavioris an unclear notion , sometimesmeaning tionality . This is because movement and sometimesmeaning something akin to intentional bodily action. In contrast, if the cognitive capacities(in all their intentional glory) are taken as our explanandum, there is no danger of overlooking the , property of intentionality , sincethe object of the game becomes in part at least, to explain in virtue of what the capacitieshave this seeminglymyste rious property. Another unfortunate tendency fostered by alternative 1 is the tendency to approach cognition in a holistic way. Intelligent behavior, whateverthat may be, is clearly the product of a complex interaction among our various 2 Chapter . cognitive capacities If we take the point of cognitive scienceresearchto be the explanation of intelligent behavior, then we are confronted with the problem of capturing this interaction right from the start. If , on the other hand, we adopt the view that I have been pushing (ie ., that the objects of our explanations are the various cognitive capacities it is far easier to ), " " adopt a divide and conquer strategy. And only the latter , I submit, . makes any methodological sense Our capacities are difficult enough to explain one by one. If we are faced with the task of describing them in interaction at the outset, the job becomescompletely daunting. ' Finally , there is also a sensein which alternative l s focus on behavior is misleading with regard to the aims of the researchframework of cognitive . scienceTo say that one is interestedin explaining an action or a piece of behavior usually means that one is interested in explaining why it occurred. For example to say that we are interested in explaining the , ' phenomenondescribedin Pylyshyn s examplewould usually be to say that we want to know why the pedestrian ran to the phone booth and dialed the numbers. That is, talk about the explanation of behavior is usually . construed as a call for reasonsor causes But this is not at all the sort of that cognitive scientists are typically after. Kosslyn for instanceis , explanation certainly not interested in why his subjectsperformed the mapscanning task. (Presumably the answer to that question is either that he paid them for participating in his experiment or that they were required to do so in order to pass the introductory psychology course they were taking .) Rather, he wants to know how they did it and in virtue of what ) (mental resources they did it . The point about the lack of why- explanations requires some qualification . When cognitive scientistsseekto explain their data they certainly are , to answer why-questions of a sort- namely, questions of the form trying " " , Why , in this experiment did such-and-such a pattern of data result? Kosslyn is certainly interestedin explaining why, when subjectsperformed the map- scanningtask, their reaction times were typically a linear function " of the " distance between points scanned on the imagined map. But although this is a why- explanation, it is very different from explaining why the subjectsperformed the map- scanning task in the first place. Furthermore , I would submit that such why- explanations of the data are important only becausethey give us a way of adjudicating among alternative answers to one of the how- or what-questions designated as the basic questions of the researchframework. The Domain and the Basic Questions Alternative 2 is The view that the domain of cognitive science the propositional attitudes has been most vociferously advocated by Fodor , as in the following passage : I assume whatwewant- andwhata successful science that cognitive oughtto give .... us is a propositionalattitude psychology That is: propositionalattitudes constitute mainsubjectmatterof thetheory thetheorypurportsto tell uswhat the : attitudesareand what sortsof thingstheydo. (1985ap. 1) , propositional What is a propositional attitude? A propositional attitude is a mental state that can be analyzed into an " attitude " component (such as perceiving " " , remembering intending) and a content component, where the content , is propositional in form (e.g., that there is milk in the refrigerator). Stateswith the sameattitude component can differ in content (for example . I can both perceive that the sky is blue and perceive that the poppies have bloomed). Conversely states with the same content can differ in , attitude (for example I can both perceiveand rememberthat there is milk , in the refrigerator). As with alternative 1, there is something right about this view. Pretheoretica conceived our cognitive capacitiesdo involve propositional , attitudes. Exercising our capacity to perceive culminates in a state of perceiving exercising our capacity to remember (recall) culminates in a , state of remembering exercising our capacity to speak a sentencebegins , with an intention to speak that sentence and so on. Furthermore, part of , explaining precisely what any given cognitive capacity is and how we . exercise may well involve providing a theory of the sort Fodor envisions it For example if the capacity to perceiveinvolves a propositional -attitude , state of perceiving then saying what that capacity is at (say) an information , -processinglevel of description will involve saying what the propositional -attitude state of perceivingcomesto at that level of description. And the latter is precisely Fodor ' s program. There are, however, a number of difficulties with alternative 2. First, even though it is true that, in some sense cognitive sciencemust give an , account of some of the propositional attitudes in order to accomplish its aims (that is, in order to answer its basic questions providing such an ), account is only a small part of what cognitive scientists want to accomplish . They also want to know what mental resourcesare required for a , I Chapter 2 person to have the capacitiesin which those propositional attitudes figure, how those cognitive capacities typically get exercised and how those , interact. Thus, to say that the domain of cognitive scienceis the capacities propositional attitudes is misleading at best. Furthermore, alternative 2 as formulated above contains no explicit ' ' that quantifier, but the phrase the propositional attitudes clearly suggests what is intended is that the domain of cognitive scienceencompass all es the propositional attitudes. There are good reasonsfor believing however, , that not all the attitudes are appropriate candidates for this role. In the first place though it may be true that any cognitive capacity (as ordinarily , conceived involves a propositional attitude , the converseis false Many . ) " a propositional attitude is not a " part of any cognitivecapacity and hence 's not uncontroversially part of cognitive science researchprogram. Searle , ( 1983 p. 4) generated the following list of states that can be intentional : states belieffear hope desirelove hate aversionliking disliking doubting wondering , , , , , , , , , , whether , elation depression ,joy , , anxiety pride remorsesorrow grief guilt, rejoicing , , , , , , , hostility affectionexpectation , acceptance , forgiveness , , , , irritation puzzlement , , , , , anger admiration contempt respectindignation intention wishing wanting , , , , , , , , , , imagining fantasy shamelust disgust animosity terror, pleasureabhorrence , . , , and aspirationamusement disappointment Note that many of theseare not, in any way, part of what constitutes the uncontroversial casesof cognitive scienceresearch In particular , at the . time, cognitive sciencehas little to say about either statesof affect present , , (joy , elation, depression remorse sorrow, anger, etc.) or statesthat figure in motivation (hope, desire aversion aspiration, etc.). In sum, then while , , , it is certainly not wildly off-baseto say that the domain of cognitive science is the propositional attitudes, such a characterization doesnot capture the focus of the field in a completely accurate way. Altemadve 3 The view that the domain of cognitive scienceis human knowledge representatio and usecan be found stated on page 6 of Gardner 1985 : I definecognitivescience a contemporaryempiricallybased as effort to answer , - particularly thoseconcerned -standing with the , long epistemological questions nature of knowledgeits componentsits sourcesits developmentand its deploymen , , , , . The Domain and the Basic Questions There are several reasons why I do not think this accurately captures the . domain of cognitive science In the first place, Gardner' s statement of this alternative is much too " " broad. Human beingsrepresentand usetheir knowledge in many ways, only someof which involve the human mind. What we know is represented in books, pictures, computer databases and so forth . Clearly, cognitive , sciencedoes not study the representationand the use of knowledge in all theseforms. Thus, at a minimum , alternative 3 must be amendedto read: The domain of cognitive scienceis human knowledge representation and usein the mind. But even with this amendment alternative 3 is not quite right , for now if it looks too narrow. Narrownessis certainly a consequence we acceptthe " standard philosophical conception of knowledge as being " factive, for cognitive sciencestudiesmore than those representationswhich are true in some sense For example it is interestedin misperception the acquisition . , , of false belief, and the use of the imagination . Perhaps then we ought to , , modify alternative 3 once more so as to get rid of the factive component and speakabout somethinglike the representationand useof beliefsrather than the representationand useof knowledge. But again the characteriza tion is too narrow. Although many of our cognitive capacities(pretheoretically conceived involve beliefs there are aspectsof thesecapacities and , , ) even other entire capacities that are of interest to cognitive scienceand , . that are not captured by speakingof the representationand useof beliefs to speak or The exerciseof any productive capacity, such as the capacity . write sentencesstarts with an intention. But intentions are not beliefs Nor , are mental images although they may certainly have beliefs associated , with them. An advocate of alternative 3 might try one more move. We have seen that talking about the representationand useof knowledge is problematic becauseknowledge is factive, and that talking about the representation and use of beliefs is too narrow becauseit introduces constraints on the ) type of propositional attitude to be studied (namely, just beliefs. But there seems be an obvious way around both of thesedifficulties. The solution to . is simply to talk about representation per se Such talk is not factive, . becausethere can be false representations And we have simply left out any mention of an attitude. But this too is off the mark. Although cognitive sciencecertainly concernsitself with mental representationsand their use , such phenomenado not constitute the domain of the field (namely, what 2 Chapter theorizing in the field is designedto explain). Rather, talk about representations and about their use belongs to that theorizing itself. In particular , representationsand their useare posited to explain such things as how we , , , , perceive understand and produce language how we remember and how we reason Therefore, they do not constitute the domain of cognitive . science they are designed to explain it . I conclude that alternative 3 ; is unsatisfactory. Alternative 4 The fourth alternative is that the domain of cognitive scienceconsists . of only those human cognitive capacitiesthat are unencapsulated To say that a capacity is unencapsulated to say that the exerciseof that capacity is is sensitiveto the goals and the general" knowledge" of the person exercising the capacity. To say, in contrast, that a capacity is encapsulatedis to say that its exercisewill always proceedin a certain way without regard to thosegoals or that general" knowledge." Note that only generalknowledge is relevant to characterizing an encapsulatedor an unencapsulatedcapacity . It is quite possiblefor an encapsulatedcapacity (such as " early vision" seems be) to have recourseto a form of " tacit " knowledge as long as that to es knowledge is available only to the process underlying that capacity. Of all the alternatives we have considered this one is, in somesensethe , , most interesting. Most people agree that the domain of a scienceought, ultimately , to be picked out on theoretical grounds. Alternative 4 attempts. in etJect to anticipate what these grounds will be. It is thus interesting , because is principled in a way that the other alternatives are not. Nevertheless it , I would like to argue that , at this point in the history of cognitive science alternative 4 ought to be rejectedalong with alternatives 1- 3. , ' in Something very like alternative 4 is suggested Pylyshyn s book Computation " and Cognition ( 1984. I say " suggested becausePylyshyn never ) quite comesout and saysdirectly and explicitly that this is what he has in mind. Nevertheless a fair bit of textual evidencepoints in this direction. , ' in My reading is most explicitly supported by somepassages Pylyshyn s final chapter whosepoint , we are told , is to raise the question of " what the term cognition applies to exactly, hence what the domain of cognitive , scienceincludes." First, on page 258, in summarizing " the ideas discussed and the claims made" in the book , Pylyshyn says the following : " All indications are that certain central aspectsof human behavior depend on what we believe and what we desire on our knowledge goals, and utili , The Domain and the Basic Questions ties and on our capacity to make inferencesfrom our beliefs and desires " and turn them into intentions to act. Although he does not explicitly say at this point that it is these aspects of behavior which constitute the domain of cognitive science it is clear that this is where his focus is. , on page 266, in taking up the question of what phenomenamight Second , . fall outside the boundaries of cognitive science Pylyshyn talks about " the semantic principles I claim characterize phenomena falling outside ." This, in itself, is open to several different readings of course ; , cognition however, what he has in mind becomes quite clear when we put this statement together with passageswhich indicate what Pylyshyn takes thesesemanticprinciples to be- namely, principles formulated not only in es terms of representationsand representation governedprocess but also in terms of such notions as knowledge, belief, desire and rationality (pp. 130 , , . 21Off) The other aspectof Computationand Cognition that supports attributing " " alternative 4 to Pylyshyn concernshis notion of functional architecture. The functional architecture of a computational systemconsistsof the basic : computational resources available in the system its primitive symbol , manipulation operations, its storage and retrieval mechanisms its control structure, and so on (p. 92). The significanceof the functional architecture, " for Pylyshyn is that it marks the point in the system(the level of aggregation , " as he calls it where we must move from semantic explanatory , ) es . principles to nonsemantic ones Mental process that make use of the functional architecture and that are defined in terms of it can always be ' described semantically or in terms of the subject s beliefs desires and , , that occur within the functional architecture cannot . es inferences Process , ) (p. 130. To explain how a primitive operation works, for example one must appeal to nonsemantic biological or physical principles. consists Given that Pylyshyn holds that the domain of cognitive science of those phenomenaexplainable in terms of semantic level principles, only and given that the functional architecture is, by definition , not so explainable , it would follow that apparently cognitive phenomena that seemto require explanation in terms of the inner workings of the functional architecture . ought not fall within the domain of cognitive science Pylyshyn of his position, and I take this does seemto accept this as a consequence to support my reading of him. For example he hypothesizesthat both the , of children to acquire a grammar of a certain sort and the human capacity Chapter 2 " , capacity to fonn knowledge-independent low -level, visual representations " are dependenton featuresof the functional architecture; primarily hence he classifiesthem as noncognitive and takes them to fall outside the , .4 proper domain of cognitive science How does alternative 4 fare as a characterization of the domain of ? cognitive science In view of the current state of the art, the answer with to descriptive considerations is clear. Many researchers currently respect believe that significant aspectsof the human perceptual capacities both - are encapsulated But such capacitiesand the theories . visual and linguistic being developed to explain them are not for that reason considered . to be outside the domain of cognitive science For example Marr and , " Nishahara' s work on " early vision (Marr 1982 is consideredby many to ) . be a paradigm of good work in cognitive science Recall, however, that , in weighing whether an assumption should be consideredpart of the researchframework of cognitive science what the , field actually does or does not subscribe to at the current time can be . overridden by nonnative considerations Pylyshyn may have this sort of in strategyin mind, for certain passages his book can be read as fonning an the view that the domain should include both encapsulated argument against and unencapsulated capacitiesand in favor of the view that it should the latter. I would fonnulate the first part of the argument as contain only follows: ' Pylysbyo s Negative Argument ( 1) Phenomena that belong to the same domain should be subject to uniform principles of explanation. (2) Nonencapsulated capacities can be accounted for only in terms of " " semantic principles of explanation. " " , (3) A semantic principle of explanation is one that refers to the beliefs desires and rationality of the organism. , 4) Encapsulated capacities can be accounted for only in terms of nonseman ( principles of explanation (either purely formal or biological). (5) The conjunction of semantic and nonsemantic principles of explanation is not uniform. capacitiesshould not be (6) Therefore, encapsulatedand nonencapsulated included in the samedomain. The Domain and the Basic Questions . This establish Pylyshyn' s negative thesis The argument for his positive es thesis then continues as follows: ' Pylyshyn s Positive Argument , (7) Ceteris paribus the domain of cognitive scienceshould include only those phenomenathat either are paradigmatically cognitive or require the same kind of principles of explanation as the paradigmatically cognitive . phenomena (8) Only the nonencapsulatedcapacitiesare paradigmatically cognitive. (9) Therefore, the domain of cognitive scienceshould include only the . nonencapsulatedcapacities Although questionscan certainly be raised regarding premises1, 7, and 8, the chiefdit1iculty with Pylyshyn' s total argument lies with his conception of the semantic(premise3). There are two levelsof description relevant " to cognitive sciencethat might be characterized as " semantic and it is , descriptionsof the important not to confusethem. There are commonsense human cognitive apparatus in term of knowledge, belief, desire rationality , , etc., and there are theoretical descriptions in terms of mental representations es and representation governed computational process . If we accept ' s somewhat nonstandard in premise 3, and , usage as stipulated Pylyshyn restrict the term ' semanticprinciple of explanation' to a principle of explanation that refers to the beliefs desires and rationality of the organism , , , 2 becomesfalse in cognitive : then the argument is in trouble because premise science non- encapsulated capacities not only can be but are , accounted for by referenceto mental representationsand representation , es governed process , and, hence by principles of explanation that are not ' . semanticin the requisite sense On the other hand if we revise Pylyshyn s , third premiseand take a semantic principle of explanation to be one that refers to content-bearing objects at either the ordinary or the theoretical level of description then we seethat the semanticjnonsemanticdistinction , -cut one another. distinction cross and the encapsulatedjnonencapsulated there can now be capacities that are encapsulatedbut that In particular , require explanation in terms of representations and representation es governed process . As it turns out, this is not merely a conceptual possibility. Indeed, a number of current theories of low -level vision (i.e. those aspectsof visual " processing that deal with problems related to the recovery of physical 2 Chapter " , properties of the visual environment [ Yuille and Ullman 1990 p. 6] ) treat such process as virtually encapsulated es and representationgoverned . to Yuille and Ullman ( 1990 p. 6), " low -level vision is believed , According to act in a manner almost independent of the domain and task." Yet theories of early vision certainly posit representations For example as . , Yuille and Ullman ( 1990 p. 10 note, Marr 's landmark 1982study posits , ) two stagesof representations the primal sketch " a symbolic representation : , obtained directly from the image by detecting and describing significant featuresin the image, such as intensity edgesand texture boundaries" and , the 2! - D sketch a representationthat describes" the surrounding surfaces , and their properties, such as their depth, surface orientation , color , and texture." Thus, revision of premise 3 leads to the falsity of premise 4. I conclude that Pylyshyn' s argument doesnot go through and, hence doesnot constitute , a good reasonfor overriding the fact that alternative 4 fails to capture the clear casesof cognitive scienceresearch . 2.2 The PropertyAaumptiODS Thus far we have ascertainedthat the domain of cognitive scienceis the human cognitive capacities Do thesecapacities have any properties that . are particularly important from the point of view of cognitive science research There are five basic general properties, on my view. Thus, we ? needto include five property assumptionsas part of our domain characterization. ( Thepoint of articulating theseproperties is not to define what a cognitive capacity is. Rather, the point is to identify a number of properties in fact shared by all or most of the cognitive capacities that , prima facie, should be explained by any adequatescientific theory of cognition .) D2 (PROPERTY ASSUMPTIONS conceived the human cognitive , ) Pretheoretica11y have a number of important properties. capacities (a) Each capacity is intentional; that is, it involves statesthat have content or are " about" something. (b) Virtually all of the capacitiesare pragmatically evaluable that is, they ; can be exercisedwith varying degreesof success . (c) When success , fully exercised each of the evaluable capacities has a certain coherence cogency or . The Domain and the Basic Questions (d) Most of the evaluablecapacitiesare reliable; that is, typically , they are exercised success (at least, to somedegree rather than unsuccess . ) fully fully e) Most of the capacities are productive that is, once a person has the ; ( capacity in question, he or she is typically in a position to manifest it in a practically unlimited number of novel ways. Let us now consider each of theseproperties in turn. 02 (a) Each of the human cognitive capacitiesis intentional; that is, it involves statesthat have content or are " about" something. In order to seein what sensethe cognitive capacitiesare intentional , it is usefulto make a distinction betweenbasicand complexcognitive capacities . A complex capacity is one that can be analyzed into subcapacities , relative to our ordinary conceptual scheme a basic capacity is one that ; cannot. Our capacity to name objects, for example is complex. There is an , answer (though a crude one) to the question how we do it . First , so the , story goes we perceiveand recognizethe object (the precise relationship is betweenthesetwo subcapacities unclear); then we think of its name: then we utter its name. In contrast, consider anyone of those subcapacities say, our capacity to utter some specific word. How do we do it? Before ' doing some scientific psychology, we haven t the slightest idea. Cognitive capacitiesare often describedin the literature as input -output functions of a special sort (Haugeland 1978 Von Eckardt Klein 1978b ; ; Cummins 1983. At the level of our ordinary conceptual scheme it is most , ) ) appropriate to talk about beginning events(or states and end events (or states. For example visually perceiving a dog takes us from something , ) like a beginning event of dog-like light rays impinging upon the retina (or , of , perhaps just the presence a dog in our field of vision) to the end state of a perceptualexperience and inferring that Socratesis mortal on the basis , of the premisesthat Socratesis a man and all men are mortal takes us from the beginning state of simply considering the premisesto the end state of reaching a judgment about the conclusion. Intentionality seemsto enter the picture as follows: Each of the basic human cognitive capacities involves some intentional state at the beginning point , at the end point , or at both points. Exercising the capacity to perceiveculminates in a state of , perceiving exercising the capacity to remember (recall) culminates in a state of remembering exercising the capacity to speak a sentencebegins , with an intention to speak that sentence and so on.5 , 2 Chapter , Roughly speaking what makesthesestatesintentional is that they have content- they are " about" something. Beyond this, there is considerable . philosophical disagreementHowever, cognitive scientistsseemto agreeon the following additional points: . The intentional states relevant to cognitive scienceare propositional attitude states Note that this is not true by definition . As Searle( 1983 p. . , 7) points out , not every intentional state has an entire proposition as intentional content. To argue for this point thus requires showing that, in eachcasewhere an intentional state seemson the basisof its expressionin ( natural language to take something other than an entire proposition as ) intentional content, either the natural -language expressionis misleading and the intentional state has an entire proposition as intentional content after all or the intentional state is not one within the purview of cognitive . science Searle( 1983 pp. 30, 41) arguesquite persuasivelythat intentional , states such as wanting and perceiving which seem to be able to take a , " simple object rather than a proposition as intentional content ( I want some apple juice " ; " I seethe tree" ), really always have full propositional content as intentional content (I want that I have some apple juice, I see that there is a tree before me). The remaining genuinecasessuch as loving , , and hating, are not currently within the purview of cognitive science , . although of course this may change . Propositional attitudes are semanticallyevaluable Beliefscan be true or . false perceptionscan be veridical or nonveridical, intentions can be carried , out or not. One of the properties Searle( 1981 1983 attributes to propositional , ) attitudes is direction of fit . Intuitively , the direction of fit of an intentional state is determined by which side is amiss (the mind or the ' " world ) if " the fit doesnt come ofT (Searle 1981 p. 207). The kind of term , we use to express the relation of satisfaction between a propositional attitude and the world dependslargely on the direction of fit of the attitude ' ' in question. ' True and 'false, according to Searle( 1981 are usedto assess ), in success representingstates of affairs in the mind-to -world direction of ' fit ; in contrast, terms such as 'fulfilled ' and ' realized are used to assess in success representingstates of affairs in the world -to -mind direction of fit . . Furthermore, what makessuchstatessemanticallyevaluableis that their content specifies(under certain aspects what Searle calls conditions of ) The Domain and the Basic Questions satisfactionthat can either be satisfiedor not , dependingon how the world " is. It is this property of intentional states which allows them to be directed " at 6 objects which do not, in fact, exist. . Propositional attitudes are " intrinsically " intentional. Objects or states with content can have this content in virtue of different sorts of " grounding" properties or relations. The notion of intrinsic intentionality taxonomizes such " grounding" properties into two groups: those whose content is determined (in part, at least) by the conscious mediation of some intentional agent and those whose content is not so determined. Objects or states whose content is determined (in part, at least) by the conscious mediation of someintentional agent are said to lack intrinsic intentionality " " , (they are, instead derived ): those whosecontent is not so determinedare said to possessit . On the standard view, intentional mental states have intrinsic intentionality , whereasexpressionsin natural languageand conventional 7 signs and symbols are derived. . Propositional attitudes have significance for their subjects Not only . does a belief or a desire involve a certain attitude toward a content, but ' both the attitude and the content make a differenceto the subjects activi ties (mental and nonmental) and hence to the subject. In other words, , , " " that are propositional attitudes are not inert : they have consequences the subject who has them. At a minimum , they typically experiencedby manifest themselvesto us at the level of consciousnessIn addition , they . are capable(under the right circumstances of influencing both the character ) of our cognitive process and our actions. es D2 (b) Virtually all of the cognitive capacitiesare pragmatically evaluable that is, they can be exercisedwith varying degreesof ; success . Chapter 2 The beginning state must be of the right kind. The end state must be of the right kind . The beginning state must bear right the sort of relation to the end state. The end state must be arrived at by meansof the right kind of causalchain. For example in the caseof veridical perception, the beginning state is a , distal perceptual stimulus and the end state is a perceptual experience . However, this is not sufficient for a processto count as veridical perception . The end state must bear the right relation to the beginning state; in particular, the perceptual experiencemust be an experienceof the perceptual stimulus and, further , must reflect the properties of that stimulus reason ably accurately. And, finally , the right sort of causal chain is required . Even if the first three conditions are satisfied if the end state is , produced in some deviant way- say, by the insertion of microelectrodes into the brain of the subject- the process does not count as veridical 8 perception. We still classify process that satisfy most but not all of these four es conditions as perception but as unsuccessful , perception. Illusion, for example , involves the right sort of beginning and end statesbut a relationship and a causalchain that have gone wrong (though not radically wrong). In contrast, where none of the conditions are satisfiedwe clearly do not have perception at all. There are also, of course things that the mind can do to which the , notions of success and failure seemto be inapplicable. Most of these(such as dreaming daydreaming fantasizing and engagingin loosely structured, , , , of , meanderingsequences thoughts) are not , at present studied by cognitive sciencealthough they certainly might be. In his popular textbook Psychology , , Henry Gleitman ( 1981 puts the point this way with respect to the ) phenomenon of thinking; Psychologists who study thinking are mainly interestedin thinking in the sense reasoning pondering or reflecting. To of , , this sense from the other senses" they refer to directedthinking, , distinguish a set of internal activities that are aimed at the solution of a problem, whether it be the discovery of a geometric proof , of the next move in a chess ' , game or of the reason why the car doesnt start. In all theseactivities, the various steps in the internal sequenceare directed and dominated by the ultimate goal, the solution of the problem." (p. 312 ) The Domain and the Basic Questions D2 (c) When exercisedsuccess , each of the evaluablecognitive fully or . capacitieshas a certain coherence cogency This coherence property is equivalent to the third condition for the successful exerciseof a cognitive capacity mentioned above. It is, however, . important enough to merit a separatediscussion If we consider the beginning and end points of any of the cognitive capacities when they are exercisedsuccess , we note that they are far fully from arbitrarily related. In fact, they typically bear a fairly well-defined relationship to one another. Moreover (and this is an important point), this relationship holds in virtue of the contents of the intentional states involved . Thus, the evaluability of our cognitive capacitiesrests on the fact that some exercisesof these capacities exhibit coherence that, in turn , ; could not be the caseunlesssomeof the statesinvolved in thesecapacities were intentional . Although it seemsto be very difficult to say what this relationship amounts to in general(that is, what it is that holds true of all the cognitive ), capacities we have fairly clear intuitions regarding what form it takes in the caseof any specific cognitive capacity. There seemto be at least two : important kinds of cases input and/ or output capacities in which either the beginning or the end state is intentional, but not both ; and purely mental capacities in which both the beginning and end states are intentional . The principal basic input and/ or output capacities are perception and intentional action. (Learning is also an important input capacity, but it is not basic; it can be broken down into various subcapacities such as , ' ) perception and memory.) On the basis of Searles ( 1983 analysis of the intentional states involved in these two capacities each capacity can be , viewed as a capacity to bring it about that the relevant intentional state is satisfied The capacity to perceive when exercisedsuccess . , fully, brings it about that the conditions of satisfaction(in the sense requirement) of the of perceptualexperienceare, in fact, satisfiedby the perceptiblestate of affairs in the world that brought that experienceabout. It does this by getting the experienceto fit the world . In contrast, the capacity to act intentionally , when exercisedsuccess , brings it about that the conditions ofsatisfacfully tion (in the senseof requirement) of the intention are, in fact, satisfied by the ensuing behavior by getting the behavior right . In other words, the coherencerelation involved is the notion of satisfaction. Chapter 2 ' Searles analysis of the satisfaction relations relevant to perception and intentional action is, actually, more complicated than this. He holds that the conditions of satisfaction relevant to perception and intentional action ' not only say something about a person s immediate environment and behavior, respectively but also say something about the causal relations , . holding betweenthe world and the relevant intentional states For example 's , on his view, the conditions of satisfaction of X seeinga yellow station " wagon are that there is a yellow station wagon in front of X and the fact that there is a yellow station wagon in front of X is causing the visual " , experience (Searle 1983 p. 61). In other words, the fact that the visual experienceis caused by the state of affairs that the visual experienceis an experienceof, is, for Searle a component of the conditions of satisfaction , . of that visual experience A similar claim is made for intentional ' action. The intentional component of intentional action, on Searlesac count, is an experienceof acting. The conditions of satisfaction of this experienceof acting are not only that there be certain bodily movement or statesof the agent but also that thesehave certain causal relations to the , of acting. experience ) Haugeland ( 1978 suggestsan alternative approach for revealing the coherence capacitiesthat involve either perception or action. In particular of , he attempts to deal with such capacitiesby comparing an appropriate descriptionof the relevant nonmental state(rather than the state itself ) with a description of the content component of the intentional state. Thus, in , considering the human capacity to perceive we are supposedto compare " " a description of what a person is looking at with a description of the 's content component of the person perceptual state. And, in considering the capacity to act intentionally , we are supposedto compare a description ' of the content component of a person s state of intending to do something . with a description of what that person actually does If we choosethe right the resulting coherencerelation will be something akin to , descriptions 9 synonymy. . Let us now turn to a consideration of the purely mental capacities The first point to notice is that many of thesecapacitiesare strongly analogous . to the casesof perception and intentional action just considered That is, " " " we have " inner perception (that is, introspection) and inner action (what Gleitman, in the quote above, had in mind when he mentioned " " directed thinking ). Here again the relation of satisfaction plays a signifi- The Domain and the Basic Questions cant role. Note , however, that in casesof directed thinking the coherence relation often involves more than satisfaction. This is so for two reasons : First, such capacitiesare intentionally imbedded; one intends to exercisea mental capacity that is, itself, intentional . Second what is intended is , even always) underspecified . usually (perhaps An example should make the point clear. In discussing memory retrieval , Gleitman ( 1981 describesa situation in which a person is doing a ) crossword puzzle and needsto find an eight-letter word forAfricanant eater' . To do this part of the puzzle, the person must exercisehis or her capacity to retrieve semanticinformation . We might analyzethe successful exerciseof this capacity as follows: Beginningstate X intends (that X recalls (what an eight-letter word for ' African anteater' is End state X recalls (that an eight-letter word for ' African anteater' is ' Aardvark ' ) , Clearly, a relation of satisfaction is involved, for X does in fact, carry out X ' s intention : X does, in fact, recall an eight-letter word for ' African anteater' . Note , however, that the coherencerelation involves more than this. In addition to satisfying the original intention, the end state fills a certain gap in the beginning state- the gap marked in the above formulation ' ' by the interrogative element what . One way to seethis clearly is to reformulate the beginning state thus: Beginning state X intends (that X recalls (that an eight-letter word for ' African anteater' is ) If we consider the propositional object of the beginning state to be something " akin to an " open sentence in logic, then the relationship betweenthe . beginning state and the end state is one of true closure The end state fills in the blank in such a way as to make the proposition true. There is more to be said, however. What makes the end-state proposition true is something . quite specific The proposition being recalled makes an identity claim; furthermore, the blank being filled in is one of the relata of that identity claim. We initially identify the intentional object only by description " ' "' (i.e., an eight-letter word for African anteater ). Once recall has been successfulhowever, we can nameit (' Aardvark '). The end-state proposition , 2 Chapter is true if what was inititally described is identical to what is eventually named. As far as I can tell , all human mental capacities can be analyzed in a similar way. Consider problem solving. Any problem can always be formulated in terms of a question. For example in the introduction to their , section on problem solving Johnson-Laird and Wason ( 1977 discuss ) , . seven sample problems, aU of which are formulated as questions For instance : Problem no. 1 You are in a hotel room in Tibet , and you want to have a wash. You go to the wash-basin and discover indecipherable symbols on the taps. Which tap do you turn on for hot water? (p. 13 ) Problem no. 2 Given a human being with an inoperable stomach tumor , and rays which destroy organic tissue at sufficient intensity, by what procedurecan one free him of the tumor by theserays and at the sametime avoid destroying the healthy tissue which surrounds it? (p. 15 ) Problemno. 3 How can one construct four equilateral triangles out of six matches whereeachside of a triangle is equal in length to the length of the , ? matches (p. 17 ) If we can formulate any problem as a question then the beginning and , end statesof the capacity to solve problems can always be describedthus: Beginning state X intends (that X thinks (that the answer to such-andsuch a question is )) End state X thinks (that the answer to such-aDd such a question is thusand-so) , (Of course any particular instantiation of this schemawould replace the ' with the name 'such and-such and 'thus-and-so' ' ' ) (but Dot the of a particular question and a particular answer respectively) . , Here is a sampling of how this analysis might apply to various other : mental capacities Deductivereasoning Beginningstate X intends (that X thinks (that, given such-aDd suchpremises the conclusion which follows logically is , ) The Domain and the Basic Questions GamepiayiD I Beginning state X intends (that X thinks (that , given such-and such a " - i.e. conducive to winning " current position in game G , the next good , - legal move is ) Imagining Beginning state X intends (that X imagines (that the front of his house looks like ) And so forth . D2 (d) Most of the evaluable capacitiesare reliable; that is, typically , ) fully (at least to somedegree rather they are exercisedsuccess than unsuccess . fully We are all acutely aware of the fact that the exerciseof human cognitive capacitiesoften goeswrong, even for adults considerednormal. We suffer perceptual illusions, forget things, make slips of the tongue, reason illogically , and so on. And , clearly, some of our capacities can be counted on more than others. On the whole, however, we do quite well when it comes to perception and production, whereas purely internal capacities such as rememberingand reasoningseemto exhibit more variability . Nevertheless , it is a fact that most of the cognitive capacities of the normal adult , (that is, an adult who by standard criteria is not considered retarded neurologically impaired, and so forth ) work remarkably well most of the time. And this is a fact that seemsto demand explanation. D2 (e) Most of the capacitiesare productive that is, once a person has ; the capacity in question, he or she is typically in a position to manifest it in a practically unlimited number of novel ways. I For example once I have the relevant capacity, I can perceivescenes have , never seenbefore, reason about problems have never thought about before , appreciate music I have never heard before, generateimages I have never imagined before rememberthings I have never rememberedbefore, , and so forth. Contemporary linguistic theory, especiallyof the Chomskian ), variety (seeChomsky 1968 has emphasizedthe productivity of the capacity . for language Chomsky, however, often has a stronger notion of productivity in mind. D2 (e) characterizesthe productivity of a cognitive 2 Chapter . capacity in terms of the psychological possibility of novelexercisesIn contrast , Chomsky claims that once a personacquiresknowledgeof a particular naturaI language he or shecan in principle understandand produce an , unbounded number of sentenceswhere the " in principle" clause is meant , to bracket resourceconstraints such as limited attention , limited memory, and limited time and where the unboundedness the capacity stemsfrom of the recursive character of syntax in natural language (For example the . , of English permits any two sentences be conjoined by 'and' to to grammar form a new sentence I prefer the weaker notion of productivity because .) it carries with it less theoretical baggage At least on Chomsky' s view, talk . about what we can do " in principle" presupposes that we can distinguish betweenaspectsof the mind underlying our competenceand empirically . aspectsof the mind underlying our performance But this assumption may be empirically faIse and it certainly does not belong to the framework of , cognitive scienceper se- especially not to the component characterizing the domain of study.10 2. 3 TheGrouping Assumption D3 (GROUPING ASSUMPTION le cognitive capacitiesof the normal, typical ) T: adult make up a theoretically coherent set of phenomena or a system . , This means that, with sufficient research it will be possible to arrive at , a set of answersto the basic questions of the researchframework that (a) constitute a unified theory and (b) are empirically and conceptually acceptable . , Pylyshyn ( 1984 p. 113 seemsto have something like assumption D3 ) in mind when he writes that " cognitive phenomena are a 'natural kind ' explainable entirely in terms of the nature of the representationsand the structure of programs running on the cognitive functional architecture." To formulate the point in terms of the notion of a " natural kind " is not quite right , however. It is worth a brief sideexcursion to seewhy this is so. Considerableattention has beendevoted to the topic of natural kinds in the recent philosophical literature. Most of it, however, has been concerned with the semanticsof natural -kind termsrather than with the kinds themselves That is, the question: ' What distinguishes natural from nonnatural . kinds? has received less attention. Everyone however, seemsto ' , agreeon the following : The Domain and the Basic Questions (a) Natural kinds either are or determine a set of actual and possible ' individuals (depending on one s view of universals. ) (b) What these individuals have in common is that they resembleeach . other or are similar to each other in some respect (c) Which kinds constitute the natural kinds and what the similarity relations are that underwrite the natural kinds are questions to be answered . ultimately by science In other words, the natural kinds will be those kinds out by predicates (though, possibly, not all - see below) which picked appear in our scientific laws and theories. , (d) Hence, if one is a realist about science natural -kind facts are objective facts. " " ); (e) Natural -kind terms are projectible predicates (Goodman 1955 hence natural -kind attributions are counterfactual supporting. For example , ' ' , is gold is a projectible predicate and a natural -kind term, whereas ' ' 'is my grandmother s favorite metal is not; the reasonfor the latter is that " there are no laws that apply to things in virtue of their being metal of my ' " 11 , ) grandmother s favorite kind (Fodor and Pylyshyn 1981 p. 146. about Within this framework of agreement however, there is disagreement , whether there are any additional constraints on something counting as a natural kind . At least three views can be discernedin the literature: ] ( 1) The liberal view (clearly held by Fodor and Pylyshyn [ 1981 ; also attributable to Quine [ 1969 ). On this view, a natural kind is any kind ] . picked out by a predicate in a scientific law of our ultimate science And " level" of scientific law will do in particular , laws from the , including, any . specialsciencesIn other words, no additional constraints are imposed on (a)- (e) above. ; ; ; (2) The explanatory-primacy view (Copi 1954 Platt 1983 Linsky 1982 de Sousa 1984. This view is more restrictive than ( 1) but less restrictive ) than (3). Here, X is a natural kind just in case (a) there exists a set of , , explananda of interest to somecommunity of scientists associatedwith X; and (b) there exists a set of properties shared by all and only instancesof X , the attribution of which (in conjunction with appropriate true auxiliary . , ) hypotheses truly explains those explananda For example let us say that biologists are interested in explaining various behavioral and anatomical " facts about tigers. Suppose further , that they identify what the " essential , genetic property of being a tiger is and that the attribution of this genetic Chapter 2 property explains (in conjunction with appropriate true auxiliary hypotheses ) the behavioral and anatomical facts in question. Then, on this view, being a tiger will constitute a natural kind . In contrast, the properties picked out by the various behavioral and anatomical facts (e.g., having a particular pattern of stripes, having particular eating habits) probably will not constitute natural kinds, even though predicatesexpressingboth kinds of properties will appear in the constituent laws of our eventual true , , tiger theory. (This is assuming of course that explanatory primacy conditions (a) and (b) do not hold independently of theseproperties.) As both Copi ( 1954 and Platt ( 1983 point out , this conception of ) ) natural kinds involves a certain amount of relativism. In particular , natural " " kinds are relative to a selected explanation space determined by the interestsof the scientistsinvolved. That is, we cannot say what property or set of properties has explanatory primacy until we have decided what " " we want to explain (that is, until we have fixed our explanation space ). Even though the explanatory primacy view restricts the classof natural kinds considerably more than the liberal view does the restriction still , rule out the possibility of natural -kind attributions in the special does not . sciencesPlatt ( 1983 p. 139 recognizesthis consequence explicitly : , ) in encountered recent A fonn of argument writingsseeks philosophical commonly or of class phenomena say that of seeings that of remembering that to establish some , - is not the membership any natural kind by invoking the fact that of " we of within themembership that class find a variability in the structuralrealisa " of thosemembers that does establish . conclusionThe the presumed not . But tions is concerned themembership of remains openthat theclass phenomena " possibility -specific variation levelnaturalkind, a kind whichallowsof ( species of some ; higher of within the lowerlevelvarieties that kind. , Presumably the reasonthis possibility remainsopen is that the two conditions for explanatory primacy may still be satisfied despite the supposed . structural variability in how seeings and rememberings are realized is, there may well be a set of explananda associatedwith seeings remembe That ( ) and a set of properties shared by all and only seeings ) (rememberings the attribution of which (in conjunction with suitable . true auxiliary hypotheses truly explains those explananda If this is the ) case then, on the explanatory-primacy view, seeings(rememberings can ) , constitute a natural kind . (3) The microstructure view (held in one form or another by Mellor 1977 , Putnam [ 1975a , Sterelny [ 1983 , and Lewis [ 1983 ). There is ] ] ] ] [ The Domain and the Basic Questions a certain amount of variation in the ways this last position has beenformulated . One version is formulated in terms of the notion of explanation, one in terms of causality, and one in terms of laws. Sincefor present purposes thesevariations make no difference I have chosena formulation in terms , of the notion of explanation, thereby, making (3) a more restrictive specification of (2). Accordingly , the microstructure view is identical to the explanatory " " primacy view except that the relevant explanation space is restricted to the causal powers of things (including their power to cause certain observations in us) and the set of properties with explanatory primacy is restricted to those that are structural in a physical, chemical, or biochemical sense On my reconstruction, the view looks like this: X is a . natural kind just in case(a) there exists a set of explanandaconcerning the causalpowersof instancesof X and (b) there existsa microstructure shared by all and only instancesof X , the attribution of which (in conjunction with . appropriate true auxiliary hypotheses truly explains those explananda ) For example water is a natural kind , on this view, becausethe chemical , theory that water is H2O explains a host of facts concerning the causal powers of water (that it producessteam at 212 F, causessugar to dissolve within a certain temperature range, etc.). In contrast to the explanatory-primacy view, the microstructure view of what a natural kind is may have the consequence that there are no (or very few) natural kinds picked out by the terms of the special sciences as , , ) , Sterelny ( 1983 notes. This is because supposedly most special science kinds are functional and, hence admit of multiple (microstructural ) realizations , : " In the natural sciencesall the membersof a kind share a structure , that can be defined physically. . . . Notoriously , the sameis not true of the predicatesof the special sciences (Sterelny 1983 p. 122 ." , ) The samesort of reasoninglies behind P. M . Churchland' s ( 1982 p. 223) , claim that " for a substantial number of contemporary philosophers, thinker is not a natural kind at all." The philosophers he has in mind are those committed to functionalism. Sincefunctionalists believe that " a thinker is just any organization of matter that instantiates a certain functional organization " which can be " realized. . . in a variety of quite different physical " " , systems and since one can exploit quite different materials governed by quite different laws in order to achievedistinct instantiations of the same abstract functional organization" ( p. 224), the conclusion follows, according to Churchland . 2 Chapter To my knowledge no functionalist has ever drawn this specific conclusion , , but that is a minor quibble. (Churchland can claim that he is attributing a belief that in fact follows from beliefs a functionalist would ' acknowledgeeven if the functionalist doesnt realizethat it so follows.) The ' ' important point is that the argument works only if natural kind is used . in its most restrictive sense If it is used in either the liberal senseor -primacy sense the premise about multiple realization the explanatory , ) provides no grounds for the conclusion, as the quote from Platt ( 1983 made clear. There may very well be sets of explananda (detailing, for , ) example the behavioral properties of intelligence that can be explained perfectly well by somefunctionalist conception of what a thinker is. We come now to the point of this whole excursion into the literature on natural kinds. Is it appropriate to say that the cognitive capacitiesconstitute a natural kind if what we really want to say is that they constitute an ? appropriate domain for scientific research There are two reasonswhy we " should answer " no to this question. First , assumption D3, in fact, makes two claims, one implicit and one explicit . It says (implicitly ) that there exists a set of answersto the basic questionsof the researchprogram that meetconditions a and b of D3. This is an onto logical claim. And it also says (explicitly) that , given sufficient . research we will be able to arrive at those answers This is an epistemo , logical claim. Substituting a natural -kind claim for a domain claim is wrong becauseclaiming that the cognitive capacitiesconstitute a natural kind (in whatever version) makes only an onto logical claim: namely, that there existsa natural kind of a certain sort. Nothing is said about our being able to discover this natural kind . Second even if we focus our attention solely on the onto logical part of , ' . assumption D3 , a natural -kind claim won t suffice Depending on how we construe natural kinds, either the claim that the cognitive capacitiesconstitute a natural kind is almost certainly falseor it does not capture what we want to say. In either case formulating the final domain-specifying assumptio , in terms of the notion of natural kinds is incorrect. To seethis last point, we must consider each of the natural -kind claims in question. For ease of discussion let us abbreviate them as NL (' L ' , ' ' ' for 'liberal '), NEP (' EP' for 'explanatory primacy ), and NMs ( M S for 'micro structure' . ) . Consider first the microstructure version, NMs Although all the facts are not yet in , it strikes me as extremely unlikely that all and only instancesof The Domain and the Basic Questions all the various kinds of human cognitive capacities (visual perception, language production, music understanding remembering etc.) will have , , someone microstructural property in common, at least at a neurophysiological level. The hard part is not finding a structural property that they all share sinceall the human cognitive capacitiesexist in virtue of structures and process in the brain, they will all share various neurophysioes logical and anatomical properties. The hard part is finding some set of neural properties that distinguishes the realization of just the human cognitive : capacities from the realization of two other classesof capacities human noncognitive capacities(such as our capacitiesto swim, jump , and ' ) sleep and nonhuman cognitive capacities(such as a chimpanzees ability to solve problems). of Despite the persuasiveness these considerations P. M . Churchland , 1982 has argued for the apparently contrary thesis that thinker is a ( ) , natural kind in a fairly restrictive sense Churchland construes a natural . kind as " a kind that figures in someone or more natural laws" (p. 223, note 1). At first blush, this sounds like the liberal view; however, the only way to make sense his argument that functionalism is committed to the thesis of that thinker is not a natural kind is to read ' natural ' here as meaning ' ' ' physical . On this reading Churchland s conception comes very close to , the microstructure view. There are, however, minor differences Insofar as . the physical laws exclude the biological ones it is narrower in scope and , ; insofar as the physical laws concern more than just the microstructure of things, it is broader. For present purposes though, these non-negligible , differencesdon' t really matter. Churchland's argument trades on certain presumedsimilarities between the notion of a living thing and the notion of a thinking thing . It goeslike this: 1. Living thing is a natural kind. " While it is no doubt true that metabolic activities can manifest themselves a nomically heterogeneous in variety of substrates governed by sundry chemical principles, or by mechanical principles, or by electromagnetic principles- it is also true that there is a deeper level of physical law that comprehends all of these casesas instances and in whose terms the fundamental activities of living matter , can be characterized The theory that achievesthis remarkable result is . non- equilibrium thermodynamics a theory of great generality whoseconcern , is with the dynamics, quality, and distribution of energy" (p. 232) To . Chapter 2 : be more precise " a living thing is a dissipative system a semi-closed , local entropic minimum , whoseinternal negativeentropy filters out further " negativeentropy from the energy flowing through it (p. 233). 2 There are important parallels betweenvital activity and cognitive activity . . Specifically: " Both the vital economy and the cognitive economy appear to maintain a fairly uniform identity or character despite being instanced in a wide variety or physical systems Both economies come in . ' . . . . Further both economies ' , continuously varying grades display a very . . . . Both economiesrequire a continuous flux strong developmental aspect of quite specifickinds of high-quality energy Both economiesrepresentthe . evolved accumulation of enormous amounts of information , and both economiesplay an essentialrole in the reproduction of this accumulated information in other individuals." (p. 234) Rather than simply concluding at this point that thinker, like living thing, is a natural kind , Churchland choosesto complicate matters somewhat . He advancesthe suggestionthat what all these parallels point to is that vital activity and cognitive activity are continuous rather than distinct. " They are both instancesof the sameform of activity , instancesthat differ . only in degree Cognitive activity , we might say, is just a high-intensity version of that of which vital activity is a comparatively low -intensity version." (p. 235) It is far from clear what to say about this argument. First , bracketing for the moment the question of its successI am not certain that Churchland , evenendsup with the claim he initially intends to make (viz., that thinker is a natural kind ). This is because as de Sousa ( 1984 p. 565 points out , , , ) one of the main demands " typically made of a well-behaved theory of natural kinds" includes the demand that natural kinds manifest sharp boundaries- that they not shadeinto each other. If this demand is taken seriously, then, insofar as cognitive activity is continuous with vital activity , cognitive activity will not constitute a natural kind . Second if the , conclusion is tantamount to a natural -kind claim, it is not clear how successful is. There is at least one reasonto be wary. From a formal point it of view, Churchland' s argument is an instance of what the logicians call an " argument from analogy" - a form of inductive argument that easily lends itself to fallacious reasoning Typically , the structure of an analogical . looks like this (Salmon 1984 p. 64): , argument The Domain and the Basic Questions 1. Objects of type X have properties F , G , H , etc. 2 Objects of type Y have properties F , G , H , etc. and Z. . 3. Therefore, objects of type X have property Z as well. how good such an argument is we must look not only at the In assessing truth or falsity (or plausibility ) of the premisesbut also at whether possession of the common properties is relevant to possessionof the inferred property . Is maintaining a fairly uniform identity despite multiple realization relevant to the existenceof an underlying shared microstructure? Is , coming in continuously varying grades displaying a strong developmental , , requiring a continuous flux of high-quality energy etc. relevant? I aspect . haveno idea. And Churchland certainly doesnot help us out on this score 's not completely some reasonswhy Churchland argument is I have suggested . persuasive In the end, however, whether we are persuadedor not does not really matter to the issue at hand. Even if it were true that the human cognitive capacitiesconstitute a natural kind in the most restrictive sensesaying that they do is not tantamount to saying that they constitute , an appropriate domain for cognitive science and it is the latter claim that ' we want to make. Supposethat Churchland s suppositions are correctthat the human cognitive capacitiesconstitute a natural kind in the sense that there exists a (microstructural ) theory true of all and only the human cognitive capacities that is capable of explaining the causal powers of . thosecapacities Still , the existenceof this stateof affairs would not guarantee us answersto the basic questions of the researchprogram of cognitive sciencethat pertain to the various, distinct kinds of cognitive capacity , (such as language understanding visual perception, and imaging) rather than to all the capacities taken together. For example knowing what , or physical properties are shared by all and only the human biological , cognitive capacitieswill not help to explain how we understand sentences . or plan actions, or generateimages And this is becauseanswering questions about the latter requiresknowing about the distinguishingfeaturesof thesesubkinds, rather than about their commonalities. A similar argument applies to NL the natural -kind claim formulated in terms of the liberal view. Supposeall we require for something to count as a natural kind is that a predicate picking out this kind appear in some lawful generalization of the field. In other words, supposethat all we need is at least one true, counterfactual-supporting generalization in which the ' ' . predicate human cognitive capacity appears Surely cognitive science Chapter 2 will be able to offer us that. In fact, severalcandidatesspring immediately ' ' to mind: ' Visual perception is a human cognitive capacity , All human ' and ' Peopleare typically cognitive capacitiesare intentional in character , ' unaware of how they exercisetheir cognitive capacities. Note, however, that when we assertthat the cognitive capacitiesconstitute an appropriate domain for scientific research we intend to say much more than that cognitive sciencewill be able to come up with generalizations like the above. We intend to say what we in fact have said: that a theoretically coherent set of answerswill be forthcoming to all the basicquestionsof the field. Thus, even the most liberal conception of a natural kind is not suitable for our fourth domain-specifyingassumption. On both the most restrictive and the most liberal conception of a natural kind , claiming that the human cognitive capacities constitute a natural . kind falls short of what we want to say for similar reasons Each of the three claims we have consideredthus far- domain-specifyingassumption 03 , NMs and NL- involves both an explanans and an explanandum. In , 03 , the explanandum is fixed by the basic questions of the researchprogram . The onto logical claim being made is a relative one: that there exists . an appropriate explanansfor this explanandum In contrast the ontological , claim made by NMsand NL is not relative. NMsassertsthe existenceof a certain kind of microstructure theory. N L assertsthe existenceof a certain set of laws. The problem is that , having started with the explanans rather than the explanandum, the posited explanans turns out to be inadequate to the desiredexplanandum (namely, answering the basic questions of the researchprogram). The situation with respect to NEP the explanatory-primacy version, is ' . in a sense the reverse Here we start with the explanandum rather than just the explanans furthermore, our choice of explanandum is unrestricted. ; Thus, we are free to choose the basic questions of the research as the . desired explanandum. The problem arises with respect to the explanans ' Recall the second requirement for somethings being a natural kind , on the explanatory-primacy view: X is a natural kind only if (b) there exists a set of properties shared by all and only instancesof X , the attribution of which (in conjunction with appropriate true auxiliary hypotheses truly ) . explains thoseexplananda It is the existenceof sucha sharedset of explanatory properties that is highly doubtful . .4 In section2 I will suggestthat among the basicquestionsof ANTCOG ' are questions of the form In virtue of what does a typical normal adult The Domain and the Basic Questions have the capacity to ?' . Supposenow that we raise this question : about each of our cognitive capacities perception, memory, language understanding languageproduction , and so on. Is it evenremotely plausible , that the cognitive capacities have in commonsome set of properties ? appeal to which would answer each of thesedistinct questions Certainly not on a cognitivist approach, for (as we shall see cognitive science ) ' attempts to answer these questions by appeal to a person s representational and computational resources But it is precisely because these . resourcesare different for each capacity that the capacitiesthemselves are different. In sum, then the situation is this: NMSeither is false or (if we accept , Churchland' s argument) doesnot say what we want to say. NL is probably true, but it also does not say what we want to say. And NEPis almost . certainly false Thus, each version either is certainly false or makes the claim. wrong Note , however, that this argument in no way impugns the existencein . generalof natural kinds in cognitive science What I have beencalling the " subkinds" of are probably natural kinds. No matter cognitive capacities what our scienceof the mind eventually looks like - that is, even if we end up with something very unlike a computational theory of mind - we will probably still have laws about visual perception, language production , remembering etc. Thus, each of the subkinds will constitute a natural , kind , at least on the liberal view. It is also quite likely that all and only instancesof each subkind will share some underlying properties sufficient for explaining most of what we want to explain about the subkind in question. If so, then each subkind will constitute a natural kind , on the explanatory-primacy view. It is even possiblethat the explanatory properties will be microstructural properties (this is certainly the assumption of neuroscience thus vindicating the microstructure version of the natural ), kinds claim for the subkinds as well. 2 The BasicQuestions .4 A researchframework getsits direction , on my account, from the questions it attempts to answer I have already suggested . that the domain of cognitive scienceconsists of the human cognitive capacities But exactly what . about human cognition do cognitive scientistswant to know? More precisely , what questions are they asking about human cognition? 2 Chapter There are four kinds of basic empirical questionsthat motivate research in cognitive science Questions of each kind can be raised for each of the . cognitive capacities included in the domain. I will expressthese kinds of questions in the form of interrogative schemasin which the blank is to be filled in with an expressionreferring to some particular cognitive capacity ' ' ' ' (such as recognizewords or perceivea scene . ) Recall that the basic empirical questions of a researchframework are anchored in our pretheoretic conception of the domain (as characterized by the domain-specifying assumptions. They are vague in the sensethat , ) for any basic question Q , it is not entirely clear what should count as a " " possibleanswerto Q . The approach adopted by a researchframework is . specifiedby its substantiveand methodological assumptions The substantive in particular , function as constraints on possibleanswers , assumptions to the basicempirical questions Because identity of any given question . the is determined by its possible answer set, in constraining the basic questions . , the substantive assumptions in effect define a new set of questions , , I shall call these the revised basic questions To put the matter slightly . differently: The questions of a researchprogram are initially posed in a " " theory-neutral fashion so that , in principle, they can be shared by competing researchframeworks and answeredin radically different ways. The role of the substantive assumptions of the researchframework is then to constrain the set of possibleanswersto those answersthat are theoretically framework (for example thosethat are framed appropriate for that research , entirely in information -processingterms). In this section, I will characterize the original basic questions of ANTCOG . The revised basic questions will be discussedin chapter 9 after I have had a chance to layout the substantiveassumptionsof the researchframework. The basicquestion schemas the researchframework of ANTCOG are of the following . Ql For the normal, typical adult , what preciselyis the human capacity to ? At the outset of the research process we can describe the cognitive , capacities of interest in a rough-and-ready manner. However, we do not know precisely what they are, either empirically or theoretically. The first question is a requestto sharpenour understandingof eachof the capacities so that questionsQ2 - Q4 can be posedmore precisely Deeper understanding . can be gained in four ways. The Domain and the Basic Questions First , there is the route of philosophical reflection on our commonsense conception. My characterizing the bulk of our cognitive capacities as intentional , pragmatically evaluable coherent, reliable, and productive is , the result of such reflection, but much more can be said along precisely theselines. For example if the object of investigation is our capacity to use , , images we can ask the following about this specific capacity: In what , respects(if at all) is it intentional , pragmatically evaluable coherent reliable , , and productive? Second cognitive scientists especially cognitive psychologists and , , - seek greater empirical understanding of our cognitive capacities linguists . How doesthe capacity manifest itself in particular circumstances For ? , example what are the temporal characteristicsof exercisinga given capacity ? Recall, for example Kosslyn's map-scanningexperiment. Information , about such temporal characteristics is not at all part of our ordinary knowledge of cognition , and it can be very revealing when it comes to answeringquestions of the form exhibited in schemasQ2- Q4. A third route for gaining deeper understanding of the capacity under study is to reconceptualize it in terms of the approach adopted by the researchframework. How , for example might we understand the human , ' capacity to usemental imagesin information -processingterms? Kosslyn s use of a CRT metaphor at the outset of his research is an example of answering Ql in this secondway. A related approach is to investigate the scope and the limits of the capacity under study. When a person fails to exercisea capacity success fully (under normal circumstances, what form does this failure take? To ) the question another way: Under what conditions do errors begin to put manifest themselves and what is the nature of theseerrors? Both Simon , ( 1981b and Pylyshyn ( 1984 have emphasizedthe importance of observing ) ) a device in atypical circumstancesin order to make inferencesabout its internal structure. In discussing a motor as an example of an artificial " , , system Simon ( 1981b p. 16 writes: In a benign environment we would ) learn from the motor only what it had beencalled upon to do; in a taxing environment we would learn something about its internal structure, specifically about those aspectsof the internal structure that were chiefly instrumental in limiting performance" It is for just such reasons that . " " cognitive scientists have attached much importance to error data in . trying to understand the cognitive capacities 2 Chapter In virtue of what does a nonnal , typical adult have the capacity to , (suchthat this capacity is intentional , pragmatically evaluable coherent, reliable, and productive)? Q2 People do not have their cognitive capacitiesmagically. There is something about the mindfbrain in virtue of which they have the capacitiesthey do. The second question is a request for a description of the mental or that make any given capacity possible in either its successful resources , fonn . Of course when this question is posedin the context its unsuccessful , of a particular research framework, researchershave a general idea of what the resourcesin question are like. Thus, to pose this question in the context of cognitive scienceis to assumethat the resourcesin question are representationaland computational . An example of answering this question about our capacity to use visual mental images from a cognitive scienceperspective can be found in Kosslyn' s general theory of mental imagery. (Seefigure 1.3 and table 1.1 above.) Q3 How doesa nonnal , typical adult typically (exercisehis or her capac . ? . Ity to) Whereas Q2 is after a static description of what underlies any given capacity, Q3 seeks a dynamic account. How are the mental resources describedin answer to Q2 actually deployed when a person exercisesthe capacity in question? What stages or steps does a person typically go fully through when the capacity is exercisedsuccess ? What exactly goes when the capacity is exercised unsuccess ? Cummins ( 1975 , fully wrong 1977 1983 provides a nice account of this sort of explanation in his , ) ." discussionsof " functional analysis Kosslyn answersthis sort of question when he develops specific dynamic models for the various subcapacities . involved in the imagery system (Seefigures 1.5 and 1.6 above.) of the nonnat, typical adult Q4 How does the capacity to ? interact with the rest of his or her cognitive capacities Thus far, each of the capacities has been treated in isolation. This question is a requestfor a more integrated account. How does our capacity to perceiveinteract with our capacitiesto remember and so on? Here , I might also introduce the distinction embodied in Q2 and Q3. That is, I might ask what it is about our mental resources that allows our capacitiesto interact (a question akin to Q2), and how our various cogni The Domain and the Basic Questions tive capacities interactwhenthey are beingexercised questionakin to (a . Q3) I have now completed my discussion of the domain and the basic . questionsof cognitive scienceThe next task- one that will occupy me for the remainder of the book - will be to get clear on how cognitive science . goesabout trying to answer those basic questions In chapters 3- 8 I will explore the conceptual and theoretical resources invoked by cognitive sciencefor this purpose. Finally , in chapter 9, we will seewhat happens when these conceptual and theoretical resourcesare combined with the basicquestionsI havejust outlined. In particular, we will seehow the basic questionsare transformed when it is assumedthat the cognitive mind is an information -processingsystem .

Find millions of documents here - Study Guides, Homework Solutions, Papers, Exam Answer Keys and more. Course Hero has millions of course related materials that will enable you to learn better, faster and get an A in all your courses.
Below is a small sample set of documents: