Nonlinear neural dynamics in the observed random variability of saccadic latencies

Behavioural reaction times to stimuli are much longer than can be justified by a simple summation of nerve conduction times and synaptic delays, and it is clear that decision processes involving higher cortical regions are involved in delaying responses. Reaction times are also unexpectedly variable from trial to trial, and their recinormal distribution has been shown to be Gaussian. In an organism faced with a complex environment filled with numerous stimuli competing for attention, this random variability in response times will result in a randomisation of attention, and consequently, of behaviour. Such randomisation of behaviour would carry evolutionary advantages in competitive situations, and it is possible that the nervous system purposefully makes use of a randomness-generating mechanism. This study investigates the possibility of nonlinear dynamical brain activity being involved in generating the random variability observed in saccadic reaction times. Electroencephalographic (EEG) measurements were taken from subjects performing a saccadic reaction time task and nonlinear measures of EEG activity were estimated (Largest Lyapunov Exponent).
A significant positive correlation between Largest Lyapunov Exponent (LLE) estimates of the EEG and saccadic measures was found, and further analysis with surrogate data has supported the reliability of the LLE results. The possibility that a nonlinear neural mechanism is the source the observed variability is tentatively supported. As saccadic latencies show a greater diurnal variability than what would be expected from looking at short-term variation and EEG measures also vary significantly over the course of the day, an alternative explanation would be that both variables are affected by fluctuations in some other physiological function.