Gating of sensory information: Joint computations of phase and amplitude data in the midbrain of the electric fish,Eigenmannia

Publisher Website
Google Scholar
Abstract
Eigenmannia is able to determine whether the electric organ discharge (EOD) of a neighbor is of higher or lower frequency than its own EOD. For small frequency differences, Df, the fish avoids jamming by shifting its frequency away from that of its neighbor. This jamming avoidance response (JAR), therefore, requires that the fish discriminate the sign of Df. The interference pattern of two EODs of similar frequency is characterized by local modulations of the instantaneous amplitude and the spatial difference of the instantaneous phase, or ‘differential phase’, of the mixed signal. When amplitude and differential phase are plotted in a two-dimensional state plane, circular graphs are obtained with a sense of rotation that reflects the sign of Df. Behavioral studies have shown that both amplitude and differential phase modulations are required for the control of the JAR. Considering two regions of the body surface, A and B, that receive strong and weak contamination by the jamming signal, respectively, rises and falls of the signal amplitude in A will be accompanied by respective advances and delays of the signal in A relative to that in B if the jamming signal is of lower frequency, i.e. if Df is negative. A plot of amplitude versus differential phase yields a clockwise sense of rotation in this case (Fig. 1). The opposite relation between amplitude and phase modulations, resulting in a counterclockwise rotation, holds for a positive Df. For the less strongly contaminated area B, however, the relation between the sign of Df and the sense of rotation is reversed, so that for a negative Df, a rise of amplitude in B will coincide with a delay of the signal in B relative to that in A. By independent experimental control of amplitude and differential-phase modulations, we explored midbrain neurons that discriminate the sense of rotations in the amplitude-phase plane. We found that these neurons achieve this discrimination by gating amplitude inputs by differentialphase information, thus exploiting the particular combinations of amplitude and differential phase that characterize a given sense of rotation (Figs. 2–4). Since the response properties of such neurons only reflect the sense of rotation, and since the same sense of rotation can be obtained for either sign of Df (depending upon the relative contamination of the receptive fields involved), individual neurons do not yet provide unambiguous information about the sign of Df. It can be shown, however, that large populations of such neurons will, nevertheless, reliably detect the correct sign of Df (Fig. 7). Response properties of these neurons offer plausible explanations for a number of earlier behavioral observations, particularly for the notion of a precise behavior controlled by a distributed system of unreliable components.