[ CogSci Summaries home | UP | email ]
http://www.jimdavies.org/summaries/

Gosselin, F. Schyns, P.(2004). A picture is worth thousands of trials: rendering the use of visual information from spiking neurons to recognition.Cognitive Science, 28, 141-146. 

@Article{GosselinSchyns2004,
  author = 	 {Frederic Gosselin},{Philippe G. Schyns},
  title = 	 {A picture is worth thousands of trials: rendering the use of visual information from spiking neurons to recognition},
  journal = 	 {Cognitive Science},
  year = 	 {2004},
  volume = 	 {28},
  pages = 	 {141--146},
  month = 	 {March}
}

Author of the summary: Mike Belanger, 2012, mikejamesbelanger@gmail.com

Cite this paper for:

More Imaging, less clarity
The authors point out that our new brain imaging techniques have not explained a number of phenomenon related to visual stimuli. For instance, how does the visual information which enters our retina (distal stimulus) affect the degree to which a response is elicited? Furthermore, how many different ways could the distal stimulus invoke a response within the cognitive system? As an example, a single-cell based study began that IT(Inferior Temporal) neurons activate during any kind of 'detail' recognition, such as faces and hands. However, this same study found that these same neurons are activated during more simple object recognition tests, and more abstract features. In other words, it is challenging to determine exactly what sort of proximal stimulus gets delegated to specific clusters of neurons. Finally, the authors call out for a 'common language' to describe all of this visual processing.
New Categories
To begin creating a 'common language' to describe visual processing, the authors propose three categories, based on how the studies obtained information. These categories are R(represented),found through reverse correlation. Reverse correlation, in this context means examining 'spikes' in neural activity slightly after certain visual stimulii are presented, or any other response (such as a keypress) after specific stimuli. In Representation, all the responses also involved discrimination between two pieces of stimuli. Another category the authors propose is (A), meaning any available brain data, although its cause and purpose have not been determined. Finally, category P(potent) data is obtained through more controlled labratory settings. This involves manipulating a single independant variable, and examining active brain regions. All responses in Potent involved identifying, or detecting change in stimulii.
Applying the Categories
The articles which are summerized in this issue are classified into either the Represented(R) information, or into the Potent(P) category, in a chart. It is unclear why the authors chose to disregard (A) in the chart. It is likely due to the fact that (A) is an inclusive category for all brain data. On the other axis, the studies are organized by the common stages each study underwent. These stages are organized by Stimulus Set, Search Space, Noise/Sample, Observer, Task, and Response. This compilation of studies helps to put an otherwise very specialized field of visual processing into perspective. It can be quickly seen for instance, that Olman and Kersten's Tasks (for the participants) included Typicality judgements. Whereas Ringach and Shapley's Tasks simply ask participants for Passive viewing. This is a good way of comparing what would otherwise be very different studies regarding visual perception.

Summary author's notes:

Back to the Cognitive Science Summaries homepage
Cognitive Science Summaries Webmaster:
JimDavies (jim@jimdavies.org)