Reader Comments

Post a new comment on this article

The combinatoric pattern appears to be insufficient -- Ernest Greene, Academic Editor

Posted by egreene on 10 Jan 2008 at 18:17 GMT

This paper bring a number of state-of-the-art tools to bear on the study of
olfactory mechanisms, and in particular the contribution of excitatory and inhibitory
sources of control over spatial and temporal patterns of activity. For those of us who
are not interested specifically in odor discrimination, the greater relevance is the
question of how information about a stable source of stimulation, i.e., the combination
of odorants that are bound to the receptors, is transformed into a representation of
the stimulus that has different properties including temporal structure.

On the basis of classical concepts of pattern recognition, one would think that the
simultaneous presence of some combination of odorants would result in activation of
dedicated channels to register each stimulus element, and with these channels
converging on a more central neuron that was designed to respond to that particular
combination. In other words, the classical view has been that recognition is based on a
combinatoric enumeration of the elements, with anatomical convergence providing the
basis for establishing that those elements are present.

In reality, however, sensory processing is more complex. Neuronal circuits in the
olfactory bulb and other brain areas actively transform combinatorial patterns before
they are integrated at higher processing stages. This transformation generates activity
patterns that are more dynamic than their input and these likely provide for memory
and recognition. The transformations of stimulus representations are thought to be
important steps in stimulus processing, but the underlying mechanisms are still
poorly understood. This paper uses a combination of approaches to examine the
neurophysiological mechanisms that underlie the encoding of odor representations.

The potential benefit of converting a stable stimulus pattern into a more complex code
that includes temporal component may allow the message to be broadcast across a
large population of memory registers. This may provide for recognition that is based
on the content of the message rather than though precise anatomical connections.