troduction Connectionismis a loosely organized researchprogram involving researchersin computer science, psychology, and in some casesneurobiology . The researchprogram has received considerableattention both in academiaand the popular press, and is sometimes touted as a radical in our of the human mind. On the breakthrough understanding " other hand, there are those who that connectionismis more than " argue nothing high tech Lockeanassociationism . In fact, the truth probably lies somewherebetween these positions. Careful study of associationistand connectionist writings reveals not only markeddifferences , but a number of fundamentalsimilarities as well. The basic idea underlying associationismcertainly is not new (it can be found in Aristotle, according to some). We begin with Thomas Hobbes, who is interested in giving an account of our train of thinking. The section contains a famous passagein which Hobbes shows how the discussionof a civil war could be causally related to someoneasking the price of a Romanpenny . The idea of the war triggers a sequenceof related or connectedideas , resulting in the seeminglyanomalous question . John Locke develops the associationistdoctrine somewhat, arguing that some ideas come to be associatedby natural connectionsholding between them while other ideas come to be associatedthrough custom (education, interests, etc.). Locke also arguesthat associationcan account for certain kinds of pathological thinking. For example, if one has a bad experiencein a particular room, one might be unableto enter the room again without thinking of the experience. This is becausethe ideas of the room and the experiencewill have become inextricably associated. David Hume proposes certain additional principles that govern the associationof ideas: resemblance, contiguity, and causeand effect. Willam Jamesgives a helpful survey of work in associationistpsychology and ad- dressestwo very important issues: the question of whether any general associative principle might underlie the proposed associationistlaws , and the question of whether neural mechanismsunderlie associationistpsychology . Jamesthus anticipates those contemporary philosopherswho take connectionismto be grounded in neural mechanisms . We begin the contemporary debate with an introduction (by JamesMcClelland , David Rumelhart, and Geoffrey Hinton) to a version of connectionismknown asparal- lel distributed processing( POP). While these writers do not make explicit referenceto the early associationistpsychologists , it is clear that they share certain fundamental views. In POP models of memory, for example, properties might be associatedwith mutually excitatory units (processors). So, if a unit representingRene Descarteswere activated, there might be a correspondingexcitation of a unit representingthe property of being a philosopher, or the property of being French. The connection strengths between units within the network are set by training the network with a general learning algorithm that may be considereda descendentof the principles first enunciated by Locke, Hume, and subsequentassociationists . 1.46 Part IV Introduction The PDP perspective stands in marked contrast to what is sometimescalled the classicaltheory of computation, in which computation consistsof fonnal operationson complex syntactic objects. For example, on the classicalview the inferencefrom the sentenceP &Q to the sentenceP is executedby a fonnal mechanismsensitive only to the syntactic fonn of P&Q. Jerry Fodor and Zenon Pylyshyn take strong exception to the PDP paradigm, suggestingthat there are severalreasons for preferring the classical theory. They argue that PDP models, by eschewingstructure -sensitiveprocess es, give up the ability to account for a number of phenomenaincluding (i) the productivity of human es (i.e., the ability to create and sentencesof linguistic process" comprehend " unboundedlength like This is the cat that ate the rat that lived in the housethat . ), ' " ' " (ii) systematicity (understanding lack likes Jill entails understanding 1illlikes Jack), (ill) compositionality (the meaning of a sentenceis a function of the meaning of its parts), and (iv) inferential coherence( inferencesHorn , e.g., P6:Q to f ). Paul Smolensky is unconvinced that these arguments pose a problem for connec- tionism. Smolenskynotes that the kinds of raised Fodor and do problems" " by Pylyshyn not connectionist treatments of soft mental es but its argue against" " process merely ability to handle hard process es suchas logical inferenceas well. Smolenskyconcedes that there are structure-sensitive es, that need not be handled process speculating they " in a classicalmodel but could be accounted for by that the mind is a supposing " statistics-sensitiveengine operating on structure-sensitivenumerical representations . Cairns and counterclaims regarding PDP systems abound today, but Seymour Papert often some deflationary remarks. He notes that the mathematicalproperties of PDP networks have yet to be explored and suggeststhat even for very simple ancestors of thesenetworks , the actual properties are difficult to detennine, and once determined , often unexpected. The abilities of full-blown connectionistsystems ( as opposed to toy implementations) are simply unknown. Further Reading Scxnedaslical worb in 88 Iodationilt psydt Ology: Hartley, D. 1749. 0b SIr N1ionson M Rn , HisfrRmI , HisDuly , midHis &", dlliions. Lo Mon. Hebb, c. O. 1949. Theo .r...imfionof Be Moior. New Yark: JohnWiley . M8\ d~ , C., - MandI , J., eds. 1964. Thinking: From A8 i RI ionto Gist.Jt. New York: JohnWiley . MilL J. 181.9. AMl .vsiI of theP I I Inom In Rof theHNmRrJ Mind . London. Scxnefurtherreadin 8 on connectionism: Bedtte L W., - A Abraham son. 1990. COr I I I Idionismmid theMind : An bttro Judionto PR1811e1 PrO Ct S6ing Net JD Orb . Oxford: BasilBlackwell . " Fodor, J., andB . Md. 1990. Connectionismand the Problemof : s aughlin' " SystematidtyWhy Smolensky' SolutionDoem t Work. Cognition35 , 183- 1.04. Minsky, M. , andS . Papert. 1986. :'5i~1" vns. Expal~ edition. Cambridge, MA: MIT Press. Rwnelhart, D., J. M~ eUand, - the pop ReseardtCroup . 1986. ParallelDistributtd Pr . ing, voL 1. Cana . , MA: MIT Press. M~ Uand, J., D. Rwnelhart, - the pop ReseardtCroup . 1986. ParallelDistributtd PrO CI SSing , voL 1.. Cambridge, MA: MIT Press. Chapter 35 Of the Consequenceor Train of Imaginations ThomasHobbes By consequence, or train of thoughts, I understand that succession of one thought to another , which is called, to distinguish it from discourse in words , mental discourse. When a man thinks on any thing whatsoever , his next thought after, is not altogether so casual as it seems to be. Not every thought to every thought succeeds indifferently . But as we have no imagination , whereof we have not formerly had sense, in whole , or in parts; so we have no transition from one imagination to another , whereof we never had the like before in our senses. The reason whereof is this. All fancies are motions within us, relics of those made in the sense: and those motions that immediately succeeded one another in the sense, continue also together after sense: insomuch as the former coming again to take place, and be predominant , the latter follows , by coherence of the matter moved , in such manner, as water upon a plane table is drawn which way anyone part of it is guided by the finger . But because in sense, to one and the same thing perceived, sometimes one thing , sometimes another succeeds, it comes to pass in time , that in the imagining of any thing , there is no certainty what we shall imagine next ; only this is certain, it shall be something that succeeded the same before, at one time or another . Train of ThoughtsUnguided This train of thoughts, or mental discourse, is of two sorts. The Ant is unguided, without design, and inconstant; wherein there is no passionatethought , to govern and direct those that follow, to itself, as the end and scope of some desire, or other passion: in which casethe thoughts are said to wander, and seemimpertinent one to another, as in a dream. Suchare commonly the thoughts of men, that are not only without company , but also without careof any thing; though even then their thoughts are asbusy asat other times, but without harmony; as the sound which a lute out of tune would yield to any man; or in tune, to one that could not play. And yet in this wild ranging of the mind, a man may oft-times perceivethe way of it, and the dependenceof one thought upon another. For in a discourse of our present civil war, what could seem more impertinent, than to ask, as one did, what was the value of a Roman penny? Yet the coherenceto me was manifest enough. For the thought of the war, introduced the thought of the delivering up the king to his enemies; the thought of that, brought in the thought of the delivering up of Christ; and that again the thought of the thirty pence, which was the price of that treason; and thenceeasily followed that maliciousquestion , and all this in a moment of time; for thought is quick. Train of Thoughts Regulated The second is more constant; as being regulated by some desire, and design. For the impression made by such things as we desire, or fear, is strong , and permanent, or, if it 248 Thomas Hobbes cease for a time, of quick return : so strong it is sometimes, as to hinder and break our sleep. From desire, arises the thought of some means we have seen produce the like of that which we aim at; and horn the thought of that , the thought of means