Influence of uncertainty and surprise on human corticospinal excitability during preparation for action

Abstract

Actions are guided by prior sensory information [1-10], which is inherently uncertain. However, how the motor system is sculpted by trial-by-trial content of current sensory information remains largely unexplored. Previous work suggests that conditional probabilities, learned under a particular context, can be used preemptively to influence the output of the motor system [11-14]. To test this we used transcranial magnetic stimulation (TMS) to read out corticospinal excitability (CSE) during preparation for action in an instructed delay task [15, 16]. We systematically varied the uncertainty about an impending action by changing the validity of the instructive visual cue. We used two information-theoretic quantities to predict changes in CSE, prior to action, on a trial-by-trial basis: entropy (average uncertainty) and surprise (the stimulus-bound information conveyed by a visual cue) [17-19]. Our data show that during preparation for action, human CSE varies according to the entropy and surprise conveyed by visual events guiding action. CSE increases on trials with low entropy about the impending action and low surprise conveyed by an event. Commensurate effects were observed in reaction times. We suggest that motor output is biased according to contextual probabilities that are represented dynamically in the brain.

Figure 1

Experimental Task and Explanatory Variables (A) Schematic of the task. On valid trials, a preparatory CS predicted the identity of a subsequent IS, cueing a button press with the right thumb or little finger. On invalid trials the CS-IS mapping was invalid as the CS was followed by the alternative IS. The validity of the CS varied across blocks of 105 trials between 85:15%, 70:30%, and 55:45%, respectively, creating blocks with, low, medium, and high uncertainty about imperative stimuli. A single TMS pulse was applied during every trial, 200 ms before IS appearance. (B) Information theoretic and categorical quantities for two experimental blocks. Examples are shown of entropy and surprise during blocks with valid-invalid CS distributions of 85:15% (left panel) and 55:45% (right panel), respectively. Top panel, entropy; middle panel, surprise; and lower panel, regressors for a categorical model containing valid and invalid trial types. The ensuing time series were used as predictors for modeling CSE and RTs across the entire series of trials of each participant.

Figure 2

Influence of Entropy and Surprise on Reaction Times and Delay Period Corticospinal Excitability (A) CSE + SD for validly and invalidly cued trials from all subjects, plotted against entropy (left) and surprise (right). For display purposes data from all subjects were binned in steps of 0.005 (entropy) and 0.025 (surprise) bits, respectively. CSE was quantified from the peak-to-peak amplitude of motor-evoked potentials (MEP), elicited in the hand muscles contralateral to the TMS stimulation site. CSE was generally higher when uncertainty (entropy) was low, and trials were preceded by surprising events. (B) RTs + SD for validly and invalidly cued trials plotted against entropy (left) and surprise (right). For display purposes data from all subjects were binned in steps of 0.005 (entropy) and 0.025 (surprise) bits, respectively. Reaction times were generally faster when uncertainty (entropy) and surprise were low.

Figure 3

Modeling Results (A) Model predictions given by the posterior densities (mean ± SD) of weights of regressors from the model containing entropy and surprise. These represent a population effect (i.e., given data and a model over all subjects). For example, for RTs, these encode response time per bit of information presented to subjects. RTs (gray bars) increased with uncertainty, Ĥ, and surprise, î. Conversely, CSE (blue bars, little finger; red bars, thumb) decreased with entropy and surprise. Responses for each muscle were modeled for trials in which the corresponding imperative stimulus had occurred. (B) Model comparison. Bar-plot of the log-marginal likelihood ratio (i.e., approximate difference between the log evidence of two competing models) was used to compare models. Importantly, this includes a model complexity term. Positive values indicate more evidence for the model containing both entropy and surprise, whereas negative values indicate more evidence for one of the alternative models. For both outcome measures (RTs, gray; CSE, blue [little finger]; red [thumb]), we found substantially more evidence (ratio >3, indicated by the horizontal black lines) for the model accounting for both uncertainty and surprise. Abbreviations: Ĥ, entropy; î, surprise; ANOVA, conventional model comprising indicator variables identifying trial type; no forgetting (all trials over all blocks are taken into account); forget₄, near maximal forgetting (only the previous four trials are taken into account).