Proceedings of The Physiological Society

University of Manchester (2010) Proc Physiol Soc 19, PC229

Poster Communications

Saccadic latency with unexpected distraction

M. Singh1, R. Carpenter1

1. Department of Physiology, Development and Neuroscience, University of Cambridge, Cambridge, United Kingdom.


  • Figure 1. Reciprobit plots of observed and simulated changes in saccadic latency distributions for two subjects when background colour was changed at the times D shown.

Reaction times essentially reflect cortical decision times: they vary randomly from trial to trial, reciprocal latency in general being Gaussian. In the LATER model this is explained by a decision signal S rising linearly in response to a stimulus until it reaches a threshold for initiating action, the rate varying normally per trial[1]; this is supported by monkey electrophysiology[2]. By interpreting S as encoding log probability, this equates to an ideal, quasi-Bayesian decision-maker, confirmed (usually with saccades due to their stereotyped nature and rapidity of data acquisition) by systematically varying such factors as expectation, information supply and urgency[3]. However, it is clearly important to determine the range of its applicability using more complex tasks. One such task is countermanding, where on some trials a stop signal occurs at a delay D after the target stimulus, that tells the subject to withhold the usual response. Behaviour is then stochastic, and the probability of successfully stopping the response is a function of D. Both this proportion, and the distribution of latencies, can be modelled successfully as a function of D by two LATER units in parallel, one being activated by the stop signal and completely cancelling the operation of the other[4]. However, often something unexpected happens that may render an impending action inappropriate, even though the subject has not been given explicit instructions. It is already known that sudden changes in the visual scene can transiently increase saccadic latency[5]. In this study we introduced unexpected changes at different times D after a target stimulus, to see how the latency distribution would be affected, and if this could be modelled by LATER units. Five volunteers, with informed consent, participated in the study, which had local ethics committee approval. In control trials, subjects made a saccade to a target presented randomly 5° to right or left of a central fixation spot, simultaneously extinguished. In test trials, randomly interleaved with controls, after a delay of 50, 70 or 90 ms the target was followed by a change: in different protocols (1000 trials each), either the original target was replaced with a bigger one, the background colour was changed, or a central target reappeared. Unlike in countermanding, an unexpected change did not cancel the impending movement, but caused a characteristic rightwards perturbation of the latency distribution (Fig. 1), the extent depending on the protocol, and the vertical position on D. These distributions could be accurately simulated (Kolmogorov-Smirnov, p<0.05) by a model with two parallel LATER units, in which the ‘stop’ unit caused only partial inhibition of the other unit, rather than complete cancellation.

Where applicable, experiments conform with Society ethical requirements