Pinker, Steven & Jacques Mehler, eds. 1988. Connections and Symbols. Cambridge MA: MIT Press.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....have shown that connectionist networks can be sensitive to the constituent structure of representations without explicit representation of that structure and without the use of explicit symbolic rules (e.g. Sect. 4) There is much more to the issue than this, however, and the early collection by Pinker Mehler (1988) is still a good introduction to the anti connectionist position. See also: Connectionist Models of Natural Language Processing. 5.3 Computability If connectionism is viewed as a fundamentally new approach to information representation and processing, then the question arises of its power ....

Pinker, Steven & Jacques Mehler, eds. 1988. Connections and Symbols. Cambridge MA: MIT Press.

....some finite limit. Nonetheless, the existence of such limits does not invalidate the more general competence theory of addition at the computational or what why level. 1 For discussion of connectionist approaches to language acquisition see, e.g. Elman (1990) MacWhinney and Leinbach (1991) Pinker and Mehler (1988), Plunkett and Marchman (1993) and Rumelhart and McClelland (1986) 2 In order to avoid confusion between the computational level of explanation and computational methods of investigation, I shall use the term what why level. Brent, Computational Language Acquisition 4 The implications of this ....

Pinker, S., & Mehler, J. (Eds.). (1988). Connections and Symbols. Cambridge, MA: MIT Press. Reprint of Cognition, 28.

....reasoning based on constraint satisfaction, and neurally realistic cognitive modeling. As connectionism has challenged the physical symbol system modeling of traditional artificial intelligence, there is an ongoing debate of the respective roles of these two approaches in explaining intelligence [13, 16]. However, in spite of the debate there are clear connections between both of the approaches. Case based reasoning is one of the areas where one can combine the ideas from these apparently diverse approaches fruitfully. What is the appeal in neural networks for case based reasoning First of all ....

S.Pinker and J.Mehler (eds.), Connections and Symbols. The MIT Press, Cambridge, England, 1988.

....hallucinations and cognitive disorganization. 2. Abnormal functional connectivity between different brain regions of schizophrenics is indicated by the advent of neuroimaging time series techniques like 15 O positron emission tomography (PET) and functional (fMRI) magnetic resonance imaging (Friston 1996). The calculation of abnormal correlation matrices confirms robust neuropsychological findings that schizophrenia reflects a failure to integrate intrinsically generated representations and concurrent perception (e.g. the Wisconsin Card Sort Task, the Stroop task and other attentional performance ....

....as binary codes. The question now is what relates these binary codes to things in the world of objects outside PTB; like that person there, or that relation between those two people there Addressing this question requires delving briefly into the origin of words and the notorious symbol grounding problem: Harnad (1996) presents a somewhat plausible account of how word meanings might be grounded in perceptual categories. There are three essential components to this Categorization hypothesis. The first concerns the formation of iconic representations. We might imagine these iconic representations as the analogue ....

In S. Pinker, & J. Mehler (eds.), Connections and Symbols. The MIT Press. Friston, K. (1996). Theoretical neurobiology and schizophrenia. British Medical Bulletin, 52 , 644--655.

....knowledge areas. Neural networks on the other hand have June 6, 1998 Page: 21 been powerful in low level tasks. However, learning multiplication tables or instances of Ohms law do not come naturally for these systems they learn the literal number relations, not the general law (Clark, 1993; Pinker and Mechler, 1988; Goldfarb, 1994) Although both symbolic and sub symbolic schools have developed elegant and highly successful techniques, neural nets and expert systems individually cannot handle the broad set of problems which living organisms must deal with. The reason is that neural network representation ....

Pinker, S. and Mechler, J. (eds), (1988), Connections and Symbols, MIT Press, MA.

.... (1988) They were followed by more experimental and theoretical work on the learning of the past tenses of English verbs (Plunkett and Marchman, 1991; Kim, Pinker, Prince, and Prasada, 1991; Marcus, Pinker, Ullman, Hollander, Rosen, and Xu, 1993; Pinker and Prince, 1991; Pinker, 1991; Prasada and Pinker, 1993; Egedi and Sproat, 1991) The most important conceptual issues that surfaced during the discussion were connected with the support that the eventual success of Rumelhart and McClelland s model of learning the past tenses of English verbs or similar improved PDP models could lend to eliminative ....

....i.e. to generalize appropriately about the regular past. Pinker and Prince, 1988, p. 124) Even more difficult to explain is a number of strange mistakes: e.g. squat 7 squakt, mail 7 membled, tour 7 toureder, mate 7 maded, brown 7 brawned, shape 7 shipt, and sip 7 sept. Prasada and Pinker (1993, p. 5) believe that the trained model failed to generalize to the correct past tense forms of these verbs due to the fact that it can represent only the phonological properties of verb stems, not their morphological structure and therefore it systematically errs on irregular sounding verbs ....

[Article contains additional citation context not shown here]

In S. Pinker & J. Mehler, (Eds.), Connections and Symbols (pp. 195-247). Cambridge, MA : MIT Press. Lapointe, S., Ling, C.X., & Matwin, M. (1993). Constructive Inductive Logic Programming.

No context found.

In S. Pinker & J. Mehler (Eds.), Connections and Symbols (pp. 3-71). Cambridge, Mass.: MIT Press. Fowler, C. (1977). Timing control in speech production. Bloomington, IN: Indiana University Linguistics Club.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC