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  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.
|Title of host publication||Connectionist Models of Learning, Development and Evolution|
|Subtitle of host publication||Proceedings of the Sixth Neural Computation and Psychology Workshop, Liège, Belgium, 16–18 September 2000|
|Editors||Robert M French, Jacques P Sougné|
|Number of pages||10|
|Publication status||Published - 2001|
|Name||Perspectives in Neural Computing|
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