Latency distributions for the reversal of optokinetic nystagmus

Abstract

Abstract: : Purpose:The time from the appearance of a target to the initiation of a saccadic eye movement (saccadic latency) varies randomly from trial to trial, with the reciprocals of saccadic latencies typically being normally distributed. This has been interpreted as resulting from a linear rise–to–threshold mechanism that decides if a target of interest is present and whose rate of rise follows a normal distribution between trials (Carpenter & Williams, Nature, 1995). It is not known, however, whether such decision mechanisms are confined to responses under volitional control. We investigated whether latencies for the reversal of optokinetic nystagmus (OKN) were distributed similarly to saccadic latencies, and therefore whether similar decision mechanisms exist in the reflex pathways governing OKN movements as in those responsible for volitionally controlled saccades. Methods:We generated OKN by having subjects stare at a computer screen (22º horizontal, 16º vertical) displaying random dots that drifted horizontally at 7.5º /s. The direction of the dots reversed at random intervals uniformly distributed between 1 and 2s. An infrared oculometer and computer system provided a continuous trace of the right eye’s horizontal movements, at a resolution of 10ms. We measured latency as the time from the direction reversal of the dots until the direction reversal of the slow phase of the OKN, in multiple runs lasting 41s each. We classified reversal direction by the final direction of the slow phase (nasal or temporal). Results:We analysed 2373 and 991 reversals for subjects A and B, respectively. Nasal and temporal reversals differed significantly for both subjects (Mann–Witney U test, p<0.001), and so we analysed nasal and temporal reversals separately. Latency distributions typically were better fit by a recinormal, rather than a normal, distribution (evidence ratio 14 and >1000 for nasal and temporal, respectively, for subject A, and >1000 for nasal, subject B: weighted least–squares fit and corrected Akaike’s Information Criterion), although one data set weakly favoured a normal distribution fit (evidence ratio 4 for temporal, subject B). Conclusions:Our results suggest that rise–to–threshold decision mechanisms underlie not only volitional responses, but also some reflex responses. For OKN, these mechanisms may be useful when deciding what to stabilise when parts of the field move at differing rates.

Keywords: nystagmus • eye movements