Neural activity in the dorsal medial superior temporal area of monkeys represents retinal error during adaptive motor learning

Abstract

To adapt to variable environments, humans regulate their behavior by modulating gains in sensory-to-motor processing. In this study, we measured a simple eye movement, the ocular following response (OFR), in monkeys to study the neuronal basis of adaptive motor learning in the visuomotor processing stream. The medial superior temporal (MST) area of the cerebral cortex is a critical site for contextual gain modulation of the OFR. However, the role of MST neurons in adaptive gain modulation of the OFR remains unknown. We adopted a velocity step-down sequence paradigm that was designed to promote adaptive gain modulation of the OFR to investigate the role of the dorsal MST (MSTd) in adaptive motor learning. In the initial learning stage, we observed a reduction in the OFR but no significant change in the "open-loop" responses for the majority of the MSTd neurons. However, in the late learning stage, some MSTd neurons exhibited significantly enhanced "closed-loop" responses in association with increases in retinal error velocity. These results indicate that the MSTd area primarily encodes visual motion, suggesting that MSTd neurons function upstream of the motor learning site to provide sensory signals to the downstream structures involved in adaptive motor learning.

Figure 1. Sample response profiles from monkey U to the velocity step-down sequence paradigm.

First ramps 80°/s downward; second ramps 0°/s. (a) Ocular response profiles. The ocular following responses (OFRs) during adaptation in the initial stage, trials 1–20, are shown in black, and those in the later stage, trials 381–400, are shown in red. From top to bottom, the graphs depict superimposed horizontal eye velocity profiles, average eye velocity, average eye acceleration (acc), and stimulus velocity (Stim vel) profiles. Arrows show the estimated time of ocular response onset, 51 ms for the initial stage of adaptation and 54 ms for the later stage. Upward deflections represent motion in the direction of the first ramp. To examine learning effects on the OFR, we quantitatively integrated the eye velocity over the time periods of 50–100 ms (magenta shaded column) and 150–200 ms (green shaded column), and calculated the changes in eye position for the early- and late-OFRs. (b) Learning effect curves of the OFRs. The moving averages of the OFRs over the time periods of 50–100 ms (early-OFRs) and 150–200 ms (late-OFRs) are shown (open circles). Arrows indicate the trial-windows when the responses started to change significantly. The red line in each panel indicates the fitted curve by the discrete model.

Figure 2. Effects of the velocity step-down paradigm on the neuronal/ocular responses.

(a) From top to bottom, the graphs depict the superimposed discharge rate of a dorsal medial superior temporal (MSTd) neuron, eye velocity (vel), and retinal error (err) velocity profiles during the initial learning stage to the velocity step-down sequence paradigm of Monkey U. The preferred stimulus of the neuron was 80°/s upward. All profiles were aligned at the stimulus onset. The initial learning stage is divided into the starting trials, consisting of trials 1–40 (thick solid lines), and the ending trials, consisting of trials 81–120 (thin solid lines). Magenta shaded columns show the time periods used to analyze the relationship between neuronal and ocular responses. Green shaded columns show the time periods used to analyze the relationship between visual motion on the retina and neuronal responses. (b–g) Moving averages. The first set of 20 trials was used to normalize the remaining sets, and shown are the time courses of the changes in the early-retinal error (b), the “open-loop” responses of the neuron (c), the early-ocular following response (OFR) (d), the late-retinal error (e), the “closed-loop” responses of the neuron (f), and the late-OFR (g). The arrow in each panel (d–g) shows the time of significant change.

Figure 3. The distributions of the modulation indices of 42 MSTd neurons at the end of initial learning (trials 101–120).

(a) The distribution of the modulation indices for the early-ocular following response (OFR). (b) The distribution of the modulation indices for the late-retinal error. (c) The distribution of the modulation indices for the neuronal responses during the “open-loop” period. (d) The distribution of the modulation indices for the neuronal responses during the “closed-loop” period. The shaded bars indicate the MSTd neurons showing significant changes.

Figure 4. The relationship between the time courses of the neuronal population averages, normalized averages of the OFRs, and retinal errors.

(a,b) The neuronal responses are plotted by open circles and thick lines (mean ± standard deviation). The ocular responses and retinal errors are plotted by dashed and solid lines, respectively. (a) The normalized average of the early-retinal errors (0–50 ms), the neuronal population average of the “open-loop” responses (40–90 ms), and the normalized average of the early-ocular following responses (OFRs) (50–100 ms) are depicted. (b) Shown are the normalized average of the late-retinal errors (100–150 ms), the neuronal population average of the “closed-loop” responses (140–190 ms), and the normalized average of the late-OFRs (150–200 ms). (c) Raw firing rate of each MSTd neuron (thin lines) and the average firing rate of the neurons (open circles) during the “open-loop” period. (d) Raw firing rate of each MSTd neuron and the average firing rate of the neurons during the “closed-loop” period. Note: In (a,b), to calculate the neuronal population average the firing rate of each neuron was normalized with the first set of 20 trials. In (c,d) the firing rates of the neurons were not normalized to calculate the average firing rate.

Figure 5. The relationship between the ratio of coefficient [velocity (vel)/acceleration (acc)] and modulation index.

The modulation index in the initial stage of motor learning (trials 101–120) plotted against the ratio of coefficients (vel/acc) for 18 MSTd neurons and the regression line.