Learning about an absent cause: Discounting and augmentation of positively and independently related causes

Frank Van Overwalle, Bert Timmermans

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Standard connectionist models of pattern completion like an auto-associator, typically fill in the activation of a missing feature with internal input from nodes that are connected to it. However, associative studies on competition between alternative causes, demonstrate that people do not always complete the activation of a missing feature, but rather actively encode it as missing whenever its presence was highly expected. Dickinson and Burke's revaluation hypothesis [4] predicts that there is always forward competition of a novel cause, but that backward competition of a known cause depends on a consistent (positive) relation with the alternative cause. This hypothesis was confirmed in several experiments. These effects cannot be explained by standard auto-associative networks, but can be accounted for by a modified auto-associative network that is able to recognize absent information as missing and provides it with negative, rather than positive activation from related nodes.
Original languageEnglish
Title of host publicationConnectionist Models of Learning, Development and Evolution
Subtitle of host publicationProceedings of the Sixth Neural Computation and Psychology Workshop, Liège, Belgium, 16–18 September 2000
EditorsRobert M French, Jacques P Sougné
PublisherSpringer
Pages219-228
Number of pages10
ISBN (Print)978-1-85233-354-6
DOIs
Publication statusPublished - 2001

Publication series

NamePerspectives in Neural Computing

Fingerprint

Learning
Neural Networks (Computer)

Cite this

Van Overwalle, F., & Timmermans, B. (2001). Learning about an absent cause: Discounting and augmentation of positively and independently related causes. In R. M. French, & J. P. Sougné (Eds.), Connectionist Models of Learning, Development and Evolution: Proceedings of the Sixth Neural Computation and Psychology Workshop, Liège, Belgium, 16–18 September 2000 (pp. 219-228). (Perspectives in Neural Computing). Springer . https://doi.org/10.1007/978-1-4471-0281-6_22

Learning about an absent cause: Discounting and augmentation of positively and independently related causes. / Van Overwalle, Frank; Timmermans, Bert.

Connectionist Models of Learning, Development and Evolution: Proceedings of the Sixth Neural Computation and Psychology Workshop, Liège, Belgium, 16–18 September 2000. ed. / Robert M French; Jacques P Sougné. Springer , 2001. p. 219-228 (Perspectives in Neural Computing).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Van Overwalle, F & Timmermans, B 2001, Learning about an absent cause: Discounting and augmentation of positively and independently related causes. in RM French & JP Sougné (eds), Connectionist Models of Learning, Development and Evolution: Proceedings of the Sixth Neural Computation and Psychology Workshop, Liège, Belgium, 16–18 September 2000. Perspectives in Neural Computing, Springer , pp. 219-228. https://doi.org/10.1007/978-1-4471-0281-6_22
Van Overwalle F, Timmermans B. Learning about an absent cause: Discounting and augmentation of positively and independently related causes. In French RM, Sougné JP, editors, Connectionist Models of Learning, Development and Evolution: Proceedings of the Sixth Neural Computation and Psychology Workshop, Liège, Belgium, 16–18 September 2000. Springer . 2001. p. 219-228. (Perspectives in Neural Computing). https://doi.org/10.1007/978-1-4471-0281-6_22
Van Overwalle, Frank ; Timmermans, Bert. / Learning about an absent cause: Discounting and augmentation of positively and independently related causes. Connectionist Models of Learning, Development and Evolution: Proceedings of the Sixth Neural Computation and Psychology Workshop, Liège, Belgium, 16–18 September 2000. editor / Robert M French ; Jacques P Sougné. Springer , 2001. pp. 219-228 (Perspectives in Neural Computing).
@inproceedings{b7235ae01aa24355a82ffa4da55e5624,
title = "Learning about an absent cause: Discounting and augmentation of positively and independently related causes",
abstract = "Standard connectionist models of pattern completion like an auto-associator, typically fill in the activation of a missing feature with internal input from nodes that are connected to it. However, associative studies on competition between alternative causes, demonstrate that people do not always complete the activation of a missing feature, but rather actively encode it as missing whenever its presence was highly expected. Dickinson and Burke's revaluation hypothesis [4] predicts that there is always forward competition of a novel cause, but that backward competition of a known cause depends on a consistent (positive) relation with the alternative cause. This hypothesis was confirmed in several experiments. These effects cannot be explained by standard auto-associative networks, but can be accounted for by a modified auto-associative network that is able to recognize absent information as missing and provides it with negative, rather than positive activation from related nodes.",
author = "{Van Overwalle}, Frank and Bert Timmermans",
year = "2001",
doi = "10.1007/978-1-4471-0281-6_22",
language = "English",
isbn = "978-1-85233-354-6",
series = "Perspectives in Neural Computing",
publisher = "Springer",
pages = "219--228",
editor = "French, {Robert M} and Sougn{\'e}, {Jacques P}",
booktitle = "Connectionist Models of Learning, Development and Evolution",

}

TY - GEN

T1 - Learning about an absent cause: Discounting and augmentation of positively and independently related causes

AU - Van Overwalle, Frank

AU - Timmermans, Bert

PY - 2001

Y1 - 2001

N2 - Standard connectionist models of pattern completion like an auto-associator, typically fill in the activation of a missing feature with internal input from nodes that are connected to it. However, associative studies on competition between alternative causes, demonstrate that people do not always complete the activation of a missing feature, but rather actively encode it as missing whenever its presence was highly expected. Dickinson and Burke's revaluation hypothesis [4] predicts that there is always forward competition of a novel cause, but that backward competition of a known cause depends on a consistent (positive) relation with the alternative cause. This hypothesis was confirmed in several experiments. These effects cannot be explained by standard auto-associative networks, but can be accounted for by a modified auto-associative network that is able to recognize absent information as missing and provides it with negative, rather than positive activation from related nodes.

AB - Standard connectionist models of pattern completion like an auto-associator, typically fill in the activation of a missing feature with internal input from nodes that are connected to it. However, associative studies on competition between alternative causes, demonstrate that people do not always complete the activation of a missing feature, but rather actively encode it as missing whenever its presence was highly expected. Dickinson and Burke's revaluation hypothesis [4] predicts that there is always forward competition of a novel cause, but that backward competition of a known cause depends on a consistent (positive) relation with the alternative cause. This hypothesis was confirmed in several experiments. These effects cannot be explained by standard auto-associative networks, but can be accounted for by a modified auto-associative network that is able to recognize absent information as missing and provides it with negative, rather than positive activation from related nodes.

U2 - 10.1007/978-1-4471-0281-6_22

DO - 10.1007/978-1-4471-0281-6_22

M3 - Conference contribution

SN - 978-1-85233-354-6

T3 - Perspectives in Neural Computing

SP - 219

EP - 228

BT - Connectionist Models of Learning, Development and Evolution

A2 - French, Robert M

A2 - Sougné, Jacques P

PB - Springer

ER -