Anticipatory movement timing using prediction and external cues

Abstract

Animals often make anticipatory movements to compensate for slow reaction times. Anticipatory movements can be timed using external, sensory cues, or by an internal prediction of when an event will occur. However, it is unknown whether external or internal cues dominate the anticipatory response when both are present. Smooth pursuit eye movements are generated by a motor system heavily influenced by anticipation. We measured pursuit to determine how its timing was influenced when both a predictable event and a visual cue were present. Monkeys tracked a moving target that appeared at a constant time relative to the onset of a fixation point. At a randomized time before target onset, the fixation point disappeared, creating a temporal "gap" that cued impending target motion. We found that the gap onset cue and prediction of target onset together determined pursuit initiation time. We also investigated whether prediction could override the gap onset cue or vice versa by manipulating target onset and, hence, the duration of time that the animal had to estimate to predict it. When target motion began earlier, the pursuit system relied more on prediction to trigger a movement, whereas the cue was more often used when the target moved later. Pursuit latency in previous trials partially accounted for this behavior. The results suggest that neither internal nor external factors dominate to control the anticipatory response and that the relative contributions vary with stimulus conditions. A model in which neuronal anticipation and fixation signals interact can explain the results.

Figure 1.

Experimental paradigm. A, Block design of the first experiment, showing three of the 10 possible fixation/gap configurations. Target onset time was constant at 750 ms. Gap duration varied from 0 to 450 ms, and fixation duration was equal to target onset minus gap duration. B, Additional block designs used in the second experiment. Target onset within a block was either 500 ms (top) or 1000 ms (bottom), with possible gap durations unchanged. For all experiments, pursuit latencies were measured with respect to the onset of the fixation interval (dashed line).

Figure 2.

Interpolation method used to detect anticipatory pursuit. The blue Xs are user-selected points on the filtered eye trace, and the sloped red line is the resulting regression fit. The short vertical line indicates where the regression line intersects the zero-velocity baseline (dotted line), and is defined as eye-movement onset.

Figure 3.

Typical pursuit traces. A, Anticipatory pursuit began during the gap interval. B, In a no-gap trial anticipatory pursuit began during the fixation interval. Note that the need for a catch-up saccade was mitigated. C, Not all trials showed anticipatory pursuit. For all panels, only part of the fixation and target motion intervals are shown.

Figure 4.

Histograms of anticipatory pursuit velocity. A, B, In both monkeys, anticipation occurred in the no-gap condition. C, D, Anticipation was more frequent when there was a gap, and also reached higher velocities. The total number of trials are noted and arrows indicate median values. Data for monkey BU are on the left and data for SA are on the right.

Figure 5.

Pursuit onset times as a function of gap onset time, for monkey BU and target onset = 750 ms. The gray triangle denotes the gap interval. The dashed line labeled “predict only” is a hypothetical regression fit with a slope equal to zero. It would result if the pursuit-onset data were determined solely by an estimate of target onset time, ignoring the gap onset cue. The second dashed line, labeled “cue only,” is another hypothetical regression, with a slope equal to one. It represents the case if pursuit onset occurred solely in response to the gap onset cue, with no attempt to predict target onset. The actual regression fit to the data (solid line, “fit to data”) lies between these two extremes, indicating that both prediction of target and response to cue influenced behavior over the course of the experiment.

Figure 6.

Pursuit onset times as a function of gap onset time. A, Data from blocks where target onset time = 500 ms. B, Blocks where target onset = 1000 ms. The monkey is BU. Other details are as in Figure 5. Note that for the early target onset time, the slope of the pursuit-onset data are closer to zero, indicating a greater tendency toward prediction. For the late target onset time, the data slope is closer to one, indicating a greater dependence on the gap onset cue.

Figure 7.

Regression slopes vary systematically with target onset time. For blocks with an early target onset, slopes are closer to zero, suggesting that prediction largely determines pursuit onset time. For blocks with a late target onset, slopes are closer to one, suggesting pursuit onset is more influenced by the visual cue of gap onset. Error bars are 95% confidence intervals.

Figure 8.

Effect of behavioral history on pursuit onset. Trials were grouped based on whether they were preceded by a short (S) or long (L) latency pursuit movement and plotted in a tree format. The rightmost nodes show mean pursuit onset for all qualifying trials. The middle nodes show the trials sorted by their immediately preceding trial, and the leftmost nodes further subdivide the data according to the two preceding trials. For example, the node labeled A(LS) represents all trials preceded by a short latency movement that was in turn preceded by a long latency movement. Note that late pursuit trials tend to follow other late pursuit trials, and early pursuit trials tend to follow other early trials. Error bars represent SE. One node in the SA tree is offset slightly for clarity.

Figure 9.

Conceptual model for pursuit initiation. The anticipation component acts as a clock that times expected target onset. The base clock rate is controlled by the readiness value. It varies randomly from trial to trial, as well as systematically according to the overall target onset time in a trial block. The fixation component responds to the onset of the gap interval and triggers the eye movement. There is an interaction between the fixation and anticipation components: if the anticipation clock is too fast, it will suppress the fixation mechanism and cause an early movement to be triggered. If the anticipation clock is slow, the fixation activity remains high, causing movements to be delayed.