Noise autocorrelation and jamming avoidance performance in pulse type electric fish

Abstract

The amplitude and the autocorrelation level of the noise affecting the interval between successive electric organ discharges were estimated in isolated fish and in socially interacting fish of the species Gymnotus carapo. Both quantities increased in the fish with the slower discharging rate of the pair during the interaction, and we aim to assess whether they have some functional implication for the efficiency of the jamming avoidance response performed by the fish having the faster discharging rate of the pair. For this purpose, the noisy variability of the intervals around its mean value was simulated using autoregressive models estimated from experimental recordings of isolated and interacting fish. The simulation was implemented using two autoregressive models, each representing one fish of the pair. The jamming avoidance response was included by adding transient interval shortenings to the train simulating the fish of the pair that discharges at a faster rate whenever the two trains were close to discharge simultaneously. The number of double coincidences (i.e., simultaneous discharges occurring in two successive firing cycles) of the two simulated trains was used to measure the efficiency of the jamming avoidance. This quantity was evaluated separately as a function of the autocorrelation level and amplitude of the simulated variability, in realizations with and without jamming avoidance response. Only if jamming avoidance response was included in the simulation have we found that (i) the number of coincidences decreased with the increasing of the autocorrelation and (ii) the increase in the amplitude determined a growth of the coincidence number at a rate that is inversely proportional to the autocorrelation level. We argue that the persistent correlations of the fish variability constitute an adaptation that improves the efficiency of transient interval shortenings as a jamming avoidance strategy. The long autocorrelation time prevents the disruption of the jamming avoidance performance due to increases in the variability amplitude.