Measuring sparseness in the brain: comment on Bowers (2009).

Department of Engineering, University of Leicester, LE17RH, England.
Psychological Review (Impact Factor: 7.97). 01/2010; 117(1):291-7. DOI: 10.1037/a0016917
Source: PubMed


Bowers challenged the common view in favor of distributed representations in psychological modeling and the main arguments given against localist and grandmother cell coding schemes. He revisited the results of several single-cell studies, arguing that they do not support distributed representations. We praise the contribution of Bowers (2009) for joining evidence from psychological modeling and neurophysiological recordings, but we disagree with several of his claims. In this comment, we argue that distinctions between distributed, localist, and grandmother cell coding can be troublesome with real data. Moreover, these distinctions seem to be lying within the same continuum, and we argue that it may be sensible to characterize coding schemes with a sparseness measure. We further argue that there may not be a unique coding scheme implemented in all brain areas and for all possible functions. In particular, current evidence suggests that the brain may use distributed codes in primary sensory areas and sparser and invariant representations in higher areas.

1 Follower
15 Reads
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The author briefly responds to a number of terminological, theoretical, and empirical issues raised in some postscripts. The goal is not to respond to each outstanding point but rather to address some comments that in my view confuse rather than clarify matters. He responds to Plaut and McClelland and Quian Quiroga and Kreiman in turn.
    Psychological Review 01/2010; 117(1):306-8. DOI:10.1037/0033-295X.117.1.306 · 7.97 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The current authors reply to a response by Bowers on a comment by the current authors on the original article. A typical problem in any discussion about grandmother cells is that there is not a general consensus about what should be called as such. Here, we discuss possible interpretations in turn and contrast them with what we find in our own data (arguably the closest experimental evidence of grandmother cells so far). A first and naïve interpretation of the term grandmother cell is that one and only one neuron encodes for one and only one concept (a face, an object, an animal, etc.). We agree with Bowers (2010) that this is a straw-man version of this idea-although some people still take this view when (incorrectly) arguing that if we would have grandmother cells then the concept of grandma would disappear if her dedicated cell dies-which clearly does not apply to our data. (PsycINFO Database Record (c) 2009 APA, all rights reserved).
    Psychological Review 01/2010; 117(1):297-9. DOI:10.1037/0033-295X.117.1.297 · 7.97 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Replies to the comments by Plaut and McClelland and Quian Quiroga and Kreiman on the authors original article both challenged my characterization of localist and distributed representations. They also challenged the biological plausibility of grandmother cells on conceptual and empirical grounds. This reply addresses these issues in turn. The premise of my argument is that grandmother cells in neuroscience are the equivalent of localist representations in psychology. When defined in this way, grandmother cells are biologically plausible, given the neuroscience to date. By contrast, the neurophysiology is shown to be inconsistent with the distributed representations often learned in existing parallel distributed processing (PDP) models, and it poses a challenge to PDP theories more generally.
    Psychological Review 01/2010; 117(1):300-6; discussion 289-90, 297-9, 306-8. DOI:10.1037/a0018047 · 7.97 Impact Factor
Show more