A neural circuit model for human sensorimotor timing

Abstract

Humans can rapidly and flexibly coordinate their movements with external stimuli. Theoretical considerations suggest that this flexibility can be understood in terms of how sensory responses reconfigure the neural circuits that control movements. However, because external stimuli can occur at unexpected times, it is unclear how the corresponding sensory inputs can be used to exert flexible control over the ongoing activity of recurrent neural circuits. Here, we tackle this problem in the domain of sensorimotor timing and develop a circuit-level model that provides insight into how the brain coordinates movement times with expected and unexpected temporal events. The model consists of two interacting modules, a motor planning module that controls movement times and a sensory anticipation module that anticipates external events. Both modules harbor a reservoir of latent dynamics and their interaction forms a control system whose output is adjusted adaptively to minimize timing errors. We show that the model's output matches human behavior in a range of tasks including time interval production, periodic production, synchronization/continuation, and Bayesian time interval reproduction. These results demonstrate how recurrent interactions in a simple and modular neural circuit could create the dynamics needed to control temporal aspects of behavior.can be viewed as a control signal that reconfigures the system's latent dynamics and allows the system to generate different outputs. However, recurrent neural network models are typically complex and do not offer the level of interpretability needed to engineer circuits that could use inputs for flexible sensorimotor coordination.Here, we tackle this question within the domain of time, asking how a simple and interpretable neural circuit could flexibly coordinate the timing of its output to external temporal events. Timing provides a prime example of sensorimotor coordination and is crucial in behaviors that demand the generation of delayed motor responses, generation of rhythmic movements with a desired tempo, and synchronization of movements to anticipated events. Early experiments found that when animals are asked to delay their motor responses, many neurons exhibit ramping activity during the delay period; i.e., activity that increases or decreases monotonically over time [21][22][23] . These experiments inspired a simple ramp-to-threshold model for action initiation. According to this model, actions are initiated reliably when the ramping activity reaches a fixed threshold 24,25 . When the threshold is fixed, flexible temporal control requires the system to adjust the rate of change, or speed, at which neural dynamics evolve. Indeed, it has been shown that neural circuits generate dynamics with a speed that has an inverse relationship to the instructed interval duration 9,20,26 .While the adjustment of speed is a common empirical observation, the mechanisms by which this is achieved is not fully understood. Recently, it was shown that flexible control of speed can be ...

Response to perturbations as a built-in feature in a mathematical model for paced finger tapping

“…It is only surprising that no published works so far in the paced finger tapping literature, up to our knowledge, deal with the issue of ill-definition of the main model variable when the sequence period is perturbed. On the other hand, models based on forced or coupled nonlinear oscillators [28,[30][31][32], traditionally classified as belonging to a "dynamical systems" approach, are naturally well defined even in the presence of perturbations to the period.…”

Order-indeterminant event-based maps for learning a beat

“…These systems typically rely on weak coupling assumptions 24 and can be analyzed by classical methods of phase response curves 25 . An-other recently introduced possibility for learning a beat is to employ a pulse-forced adaptable competition system consisting of units that ramp up at adjustable rates to a fixed threshold to produce specific time intervals 26 . This approach follows the error-correction paradigm in that the ramping rate is subject to a learning rule.…”

Bridging Motor and Cognitive Control: It’s About Time!

“…This study provides evidence that DMFC mediates the influence of prior predictions and incoming sensory evidence on planned actions, and lays the groundwork for critical tests of this proposed mechanism using causal manipulations (i.e., stimulation or inactivation). Such causal tests can also help to rule out alternative accounts of neural dynamics during the sample intervals, for instance, whether they reflect a simulated motor plan (as the authors infer) or an interval expectation (e.g., predicting the onset of the interval cue [8]). Nevertheless, by elaborating on the neuronal dynamics within DMFC during a task that requires online adjustments of learning and control, this study builds on a growing literature that implicates regions along this dorsomedial wall in the control of motor and cognitive commands [9,10].…”