TMS Over V5 Disrupts Motion Prediction

TLDR

Predictive coding posits that higher‑level areas generate models communicated via feedback to early sensory regions, enabling efficient processing of abundant sensory input. The study tested whether area V5 is necessary for predictive processing in an apparent motion paradigm. Targets were flashed on the apparent motion trace either in‑time or out‑of‑time with the predicted motion token, and double‑pulse TMS was applied over left V5 or vertex before target onset. Predictable in‑time targets were better detected than unpredictable out‑of‑time targets, but TMS over V5 eliminated this advantage, whereas TMS over vertex did not; thus V5 is causally required for motion prediction, supporting predictive coding.

Abstract

Given the vast amount of sensory information the brain has to deal with, predicting some of this information based on the current context is a resource-efficient strategy. The framework of predictive coding states that higher-level brain areas generate a predictive model to be communicated via feedback connections to early sensory areas. Here, we directly tested the necessity of a higher-level visual area, V5, in this predictive processing in the context of an apparent motion paradigm. We flashed targets on the apparent motion trace in-time or out-of-time with the predicted illusory motion token. As in previous studies, we found that predictable in-time targets were better detected than unpredictable out-of-time targets. However, when we applied functional magnetic resonance imaging-guided, double-pulse transcranial magnetic stimulation (TMS) over left V5 at 13-53 ms before target onset, the detection advantage of in-time targets was eliminated; this was not the case when TMS was applied over the vertex. Our results are causal evidence that V5 is necessary for a prediction effect, which has been shown to modulate V1 activity (Alink et al. 2010). Thus, our findings suggest that information processing between V5 and V1 is crucial for visual motion prediction, providing experimental support for the predictive coding framework.