next up previous
Next: Feature extraction in multiple Up: No Title Previous: Inhibition in the simple

Molecular recognition by distributed temporal and spatial patterning
in the olfactory system;
from biology to an artificial nose

John S. Kauer

Tufts Medical School

An attractive hypothesis, encompassing many experimental data, that explains how odor stimuli may be encoded by the vertebrate olfactory pathway holds that the high selectivity and sensitivity of the system arises from distributed receptor and central events that act together. We have referred to this process as 'distributed specificity'. This idea has arisen from observations that single primary olfactory neurons (OSNs) are broadly responsive to a number of chemical compounds; that responses to any one odorant are found over large regions of the olfactory epithelium; that neurons expressing message for any one putative molecular receptor are widely distributed in the nose; that OSNs responsive to one odorant and that express message for one receptor have convergent projections onto restricted olfactory bulb glomeruli; that there are widely distributed responses to one odorant across the olfactory bulb; and that the system is resistant to large central lesions indicating that odorant representation is not highly localized. These features suggest that information about a single odorant compound is encoded by activity widely dispersed in space and time across multiple levels in the system. By exploiting the findings from a number of these observations we have constructed an artificial olfactory system with sensors that are cross-reactive and integrative circuits that are based on those of the olfactory bulb. This system is capable of discriminating single carbon atom differences within homologous series of odorants. The relationships between biological and artificial olfactory systems will be discussed.

Tony Zador
Wed Mar 12 22:07:02 PST 1997