Biasing influences on the motor system during action preparation: a multimodal neuroimaging-computationally informed approach

In almost every aspect of life, adequate behavior first requires the processing of incoming sensory information. We then must weigh its relevance, decide between alternative movements, and select the most appropriate behavior. For example, a goalkeeper facing a penalty kick needs to observe player and ball, estimate the likely trajectory of the ball based on experience and postural cues from the player, and finally decide in which direction to move as fast as possible. We therefore use our experience for preparing the appropriate movements, yet it is still poorly understood how information such as past experience gets funnelled into our actual movements. Previous research suggests that the processes underlying perception and movement overlap in the brain, both anatomically and in time. However, the brain activity patterns, rules and computations exchanged by different brain regions to accomplish this influence remain largely undetermined. The central goal of my proposal is to reveal how different brain regions interact with the motor system to use past experience and prior information for movement selection. In my experiments, participants prepare movements based on visual cues. These cues indicate which movement is required (e.g. button presses) when a subsequent visual go-signal is presented. When cues are reliably predicting the required movement, responses are fast. This suggests that we use our experience about the reliability of visual information available to us to prepare movements efficiently. By changing the validity of these visual cues, one can change the uncertainty about which movement will have to be performed. In this way, we can address the question how our experience about the world influences our actions, and our brain. First, non-invasive brain stimulation will measure the activation state of the motor system during movement preparation in these experiments. This safe method for causing brain activity provides a direct read-out of the state of a brain region. I will then combine this brain stimulation with neuroimaging. I have recently developed this combined technique which allows for measuring the impact of brain stimulation on activity across the entire brain with high anatomical precision. Brain regions involved in preparing movements will be stimulated. At the same time, neuroimaging will measure the resulting activity changes in connected brain regions. This will reveal the brain regions relevant for making movements based on sensory experience, and disclose the influences among them which are required to bind together our experience with our actions. Finally, Magnetoencephalography non-invasively detects electrical brain signals in humans, and will reveal the timecourse of brain activity during preparation for movements based on experience and past information. It may seem obvious that the brain uses past information to guide our future movements. However, few studies have formally tried to quantify how our experience about sensory information shapes the motor system over time. One can quantify this experience by using simple mathematical models. These provide so-called computationally informed time-evolving representations of experience, and can be applied to test for corresponding brain activity changes. Using such computationally informed representations, for example, about the uncertainty of visual information, is exciting because it overcomes one of the central limitations that has bedevilled cognitive psychology for the past 100 years: the actual computations of the brain are hidden to us, and only by formal representation can we understand the processes that the brain embellishes for specifying movements efficiently in an uncertain world.

Decisions for actions are often guided by sensory input and past experiences, and we must learn from experience to specify and prepare actions in advance of an event. A central question in cognitive neuroscience is how we use past experience to implement actions at the level of the motor system? Recent findings show that perceptual inference, learning, and action selection occur in parallel, overlapping topographically and in time across the brain. Accordingly, new theories predict that the motor system is continuously biased by the affordances of the present situation, such as prior information, perceptual integration, or reward, to form decisions for actions. However, these predictions have not been tested in humans. The causal interplay and state-dependence of inter-cortical interactions underlying biasing influences are largely unknown. The proposed research will measure causal functional influences among brain regions during preparation of actions using a multimodal neuroimaging approach. Transcranial magnetic stimulation (TMS) and Magnetoencephalography (MEG) will test for the temporal organisation of causal influences from association cortex to M1 during action preparation. Concurrent TMS-functional magnetic resonance imaging (fMRI) will reveal the functional neuroanatomy of causal inter-regional influences during motor preparation. These experimental approaches will be complemented by computationally informed quantifications of the time-evolving (rather than average) changes of visual information. Using computational models such as information theory closes an important gap: they represent the dynamic, not just average, evolution of how experience or prior information might bias be represented in an action-dependent way. Together, the proposal will address how different brain regions interact with the motor system on a trial-by-trial basis, to transform prior information and experience into actions.