Neural Networks Learn Highly Selective Representations in Order to Overcome the Superposition Catastrophe.

A key insight from 50 years of neurophysiology is that some neurons in cortex respond to information in a highly selective manner. Why is this? We argue that selective representations support the coactivation of multiple "things" (e.g., words, objects, faces) in short-term memory, whereas nonselecti...

Descripción completa

Detalles Bibliográficos
Publicado en:Psychological Review Vol. 121; no. 2; pp. 248 - 262
Autores principales: Bowers, Jeffrey S., Vankov, Ivan I., Damian, Markus F., Davis, Colin J.
Formato: Artículo
Publicado: American Psychological Association Apr2014
Materias:
Acceso en línea:Ver este registro en EBSCOhost
fields @attributes:
  recordID: 1
pdfLink:
plink: https://search.ebscohost.com/login.aspx?direct=true&db=ssf&AN=95587734&site=ehost-live
header:
  @attributes:
    shortDbName: ssf
    uiTerm: 95587734
    longDbName: Social Sciences Full Text (H.W. Wilson)
    uiTag: AN
  controlInfo:
    bkinfo:
    jinfo:
      jid:
        0033295X
        PYV
      jtl: Psychological Review
      issn: 0033295X
      maglogo: N
    pubinfo:
      dt: Apr2014
      vid: 121
      iid: 2
      pid: 34
      pub: American Psychological Association
    artinfo:
      ui:
        95587734
        10.1037/a0035943
      ppf: 248
      ppct: 14
      formats:
      tig:
        atl: Neural Networks Learn Highly Selective Representations in Order to Overcome the Superposition Catastrophe.
      aug:
        au:
          Bowers, Jeffrey S.
          Vankov, Ivan I.
          Damian, Markus F.
          Davis, Colin J.
        affil: University of Bristol
      su:
        Neurophysiology
        Neurons
        Short-term memory
        Cerebral cortex
        Biological neural networks
      sug:
        subj:
          Neurophysiology
          Neurons
          Short-term memory
          Cerebral cortex
          Biological neural networks
      keyword:
        distributed representations
        grandmother cells
        localist representations
        short-term memory
        superposition catastrophe
        distributed representations
        grandmother cells
        localist representations
        short-term memory
        superposition catastrophe
      ab: A key insight from 50 years of neurophysiology is that some neurons in cortex respond to information in a highly selective manner. Why is this? We argue that selective representations support the coactivation of multiple "things" (e.g., words, objects, faces) in short-term memory, whereas nonselective codes are often unsuitable for this purpose. That is, the coactivation of nonselective codes often results in a blend pattern that is ambiguous; the so-called superposition catastrophe. We show that a recurrent parallel distributed processing network trained to code for multiple words at the same time over the same set of units learns localist letter and word codes, and the number of localist codes scales with the level of the superposition. Given that many cortical systems are required to coactivate multiple things in short-term memory, we suggest that the superposition constraint plays a role in explaining the existence of selective codes in cortex.
      pubtype: Academic Journal
      doctype: Article
      src: R
    language: English
    refInfo:
    copyright:
      @attributes:
        flag: N
    holdings:
      @attributes:
        islocal: N