Phillips, S., & Halford, G. S. (1997). Systematicity: Psychological evidence with connectionist implications. In M. G. Shafto & P. Langley (Eds.), Proceedings of the Nineteenth Annual Conference of the Cognitive Science Society, pp. 614--619.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....van Gelder and Niklasson, 1994; Matthews, 1994) It is, however, generally accepted by connectionists that human thought to some degree obeys the socalled compositionality and systematicity principles. In addition, further evidence of systematicity in human subjects has recently been supplied by Phillips and Halford (1997) and Halford et al. 1998) Thus, much research has focused on explicitly achieving these principles in connectionist hardware, and also supplying connectionist explanations to them (Smolensky, 1987; Smolensky, 1990; van Gelder, 1990; Pollack, 1990; Chalmers, 1990; Niklasson and Sharkey, 1992; ....

Phillips, S. and Halford, G. S. (1997). Systematicity: Psychological evidence with connectionist implications. In Proceedings of the Nineteenth annual meeting of the Cognitive Science Society, pages 614-619, New Jersey. Lawrence Erlbaum.

....van Gelder and Niklasson, 1994; Matthews, 1994) It is, however, generally accepted by connectionists that human thought to some degree obeys the socalled compositionality and systematicity principles. In addition, further evidence of systematicity in human subjects has recently been supplied by Phillips and Halford (1997) and Halford et al. 1998) Thus, much research has focused on explicitly achieving these principles in connectionist hardware, and also supplying connectionist explanations to them (Smolensky, 1987; Smolensky, 1990; van Gelder, 1990; Pollack, 1990; Chalmers, 1990; Niklasson and Sharkey, 1992; ....

Phillips, S. and Halford, G. S. (1997). Systematicity: Psychological evidence with connectionist implications. In Proceedings of the Nineteenth annual meeting of the Cognitive Science Society, pages 614--619, New Jersey. Lawrence Erlbaum.

....systematicity may merely reflect interference from other sources. Under controlled conditions, subjects consistently make inferences implied by the underlying logical rules (Halford, Bain, Maybery, Andrews, 1998) Indeed, such tasks are ideal tests for 16 systematicity in connectionist networks (Phillips Halford, 1997; Phillips, 1999) See, also, Bringsjorg, Noel, Bringsjord, 1998, who showed that such logical errors are overcome with education, and therefore do not imply cognitive limitations. The assumption that subjects can make the stated inferences for transverse patterning is reasonable when one ....

....task elements. As such, this sort of architecture is derived from what could be called relation approximation. The properties of relational knowledge have been defined in Halford, Wilson, and Phillips (1998, sect. 2. 2) and some of their implications for connectionist models have been analyzed in Phillips and Halford (1997) and Phillips (1999) 14 The problem with first order connectionism is not that it relies on similarity, but that it relies on constituent similarity. DISCUSSION Within the same task paradigm (context constituent to constituent inference) the constituent similarity proposal has gone from ....

Phillips, S., & Halford, G. S. (1997). Systematicity: Psychological evidence with connectionist implications. In M. G. Shafto & P. Langley (Eds.), Proceedings of the Nineteenth Annual Conference of the Cognitive Science Society, pp. 614--619.

....of generalization: An error surface view Steven Phillips stevep etl.go. jp Information Science Division, Electrotechnical Laboratory, Tsukuba, Japan Abstract: Feedforward networks with shared weights transfer learning to isomorphic tasks, but the degree of transfer is not the same as humans [3]. As an addendum to [3] error surface plots clarify the problem: the global minimum for training cannot be constrained to coincide with the test set minimum, hence generalization cannot be guaranteed. Such networks are rejected as the mechanism for transfer in human cognition. Introduction ....

....An error surface view Steven Phillips stevep etl.go. jp Information Science Division, Electrotechnical Laboratory, Tsukuba, Japan Abstract: Feedforward networks with shared weights transfer learning to isomorphic tasks, but the degree of transfer is not the same as humans [3] As an addendum to [3], error surface plots clarify the problem: the global minimum for training cannot be constrained to coincide with the test set minimum, hence generalization cannot be guaranteed. Such networks are rejected as the mechanism for transfer in human cognition. Introduction Apparent demonstrations of ....

[Article contains additional citation context not shown here]

S Phillips and G S Halford. Systematicity: Psychological evidence with connectionist implications. In M G Shafto and P Langley, editors, Proceedings of the Nineteenth Annual Conference of the Cognitive Science Society, pages 614--619, 1997. http://www.etl.go.jp/etl/ninchi/stevep@etl.go.jp.



....to novel position) in that one of the four strings does not appear anywhere in the first two mappings 11 . Thus, the implications of systematicity for a (connectionist) cognitive architecture cannot be so easily dismissed on the grounds that human cognition is, at best, quasi systematic (Phillips Halford, 1997). Although subjects are highly unlikely to have experience at this particular problem, knowledge of spatial analogues would greatly facilitate performance. However, that subjects can call upon such knowledge acquired through the course of development does not alter the point. Rather, it reiterates ....

Phillips, S., & Halford, G. S. (1997). Systematicity: Psychological evidence with connectionist implications. In M. G. Shafto & P. Langley (Eds.), Proceedings of the Nineteenth Annual Conference of the Cognitive Science Society, pp. 614--619.

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