Connectionist modelling in psychology: A localist manifesto

By Mike Page. 2000

299 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Generalization Through the Recurrent Interaction of Episodic Memories (2012, methodology): “…REMERGE model, therefore, and indeed the wider class of interactive activation networks (McClelland, 1981; McClelland & Rumelhart, 1981; also see Page, 2000), the appropriate level of inhibitory competition plays a critical role in allowing localist codes for several conjunctive experiences to…”
A neural theory of visual attention: bridging cognition and neurophysiology. (2005, background): “Feature i can be a more or less simple physical feature or a microfeature in a distributed representation (cf. Hinton, McClelland, & Rumelhart, 1986; Page, 2000), and a feature-i neuron may be broadly sensitive to feature i’s degree of presence rather than being sharply tuned to feature i.”
Neurosymbolic AI: the 3rd wave (2020, background): “In Pearl (2019), Pearl proposes a hierarchy consisting of three levels: association, intervention and counterfactual reasoning, and claims that ML is only capable of achieving association.”
Similarity Analysis of Contextual Word Representation Models (2020, background): “Questions of localization and distributivity of information have been under investigation for a long time in the connectionist cognitive science literature (Page, 2000; Bowers, 2002; Gayler and Levy, 2011).”
The spatial coding model of visual word identification. (2010, background): “Nevertheless, a localist account is the most straightforward means of implementing a lexicalist view (for discussion of theoretical arguments favouring localist over distributed representations, see Bowers, 2002; Bowers, Damian, & Davis, 2009, submitted; Davis, 1999; Page, 2000).”
Doing without schema hierarchies: a recurrent connectionist approach to normal and impaired routine sequential action. (2004, background): “Nevertheless, it is important to acknowledge that the issues involved in this area are the subject of current debate (Page, 2000).”
Grandmother cells and localist representations: a review of current thinking (2017, background): “reason for learning localist codes in cortex (Bowers, 2002; Page, 2000), just as catastrophic interference provided a pressure to learn selective and sparse codes in the hippocampus (Marr, 1971).”
Self-refreshing memory in artificial neural networks: learning temporal sequences without catastrophic forgetting (2004, methodology): “A reverberating feedforward network (RFN) is a standard three-layer backpropagation network that integrates an autoassociative processing constraint (Ans and Rousset 1997, 2000).”
On the biological plausibility of grandmother cells: implications for neural network theories in psychology and neuroscience. (2009, background): “The obvious implication is that the brain might also use grandmother cells for at least some purposes (cf., Bowers, 2002; Bowers, Damian, & Davis, 2008; Gardner-Medwin & Barlow, 2001; Page, 2000, for some computational reasons why a PDP model might learn localist representations).”
A Connectionist Cognitive Model for Temporal Synchronisation and Learning (2007, background): “Although challenging, the construction of such computational cognitive models would meet the requirements for a long standing problem in artificial intelligence: the integration of the connectionist and the symbolic paradigms of artificial intelligence, which has long been recognised as a standing r”
Neural-Symbolic Learning and Reasoning: A Survey and Interpretation (2017, background): “Although not always explicitly framed in terms of compositionality, the intermittent attacks on PDP approaches by proponents of localist neural coding schemes have a similar flavour (Page, 2000).”
The architecture of the bilingual word recognition system: From identification to decision (2002, background): “More generally, both IA and PDP models appear to have their strengths and weaknesses (see Grainger and Jacobs, 1998; Page, 2000 for the advantages of IA models).”
Merging information in speech recognition: Feedback is never necessary (2000, background): “…long-term average behaviour of a connectionist network follow the Luce choice rule by adding noise and then selecting the node with the largest activation (Page 2000; McClelland 1991), there is no such obvious way to derive response probabilities directly from network activations on a single trial.”
The Bayesian reader: explaining word recognition as an optimal Bayesian decision process. (2006, background): “In particular, localist models are capable of single trial learning (see Page, 2000, for a discussion of the merits of localist connectionist models).”
Semantic Cognition: A Parallel Distributed Processing Approach (2004, background): “2It may be worth noting that some, but not all, of the properties seen in the Rumelhart network also arise in localist networks that make use of bidirectional propagation of activation (Page, 2000).”