Dissociating dynamic probability and predictability in observed actions-an fMRI study.

Christiane Ahlheim,Waltraud Stadler,Ricarda I Schubotz

doi:10.3389/fnhum.2014.00273

Abstract

The present fMRI study investigated whether human observers spontaneously exploit the statistical structure underlying continuous action sequences. In particular, we tested whether two different statistical properties can be distinguished with regard to their neural correlates: an action step's predictability and its probability. To assess these properties we used measures from information theory. Predictability of action steps was operationalized by its inverse, conditional entropy, which combines the number of possible action steps with their respective probabilities. Probability of action steps was assessed using conditional surprisal, which increases with decreasing probability. Participants were trained in an action observation paradigm with video clips showing sequences of 9–33 s length with varying numbers of action steps that were statistically structured according to a Markov chain. Behavioral tests revealed that participants implicitly learned this statistical structure, showing that humans are sensitive toward these probabilistic regularities. Surprisal (lower probability) enhanced the BOLD signal in the anterior intraparietal sulcus. In contrast, high conditional entropy, i.e., low predictability, was correlated with higher activity in dorsomedial prefrontal cortex, orbitofrontal gyrus, and posterior intraparietal sulcus. Furthermore, we found a correlation between the anterior hippocampus' response to conditional entropy with the extent of learning, such that the more participants had learnt the structure, the greater the magnitude of hippocampus activation in response to conditional entropy. Findings show that two aspects of predictions can be dissociated: an action's predictability is reflected in a top-down modulation of attentional focus, evident in increased fronto-parietal activation. In contrast, an action's probability depends on the identity of the stimulus itself, resulting in bottom-up driven processing costs in the parietal cortex.

Highlights

When we observe another person’s action, we are quite accurate at predicting what is going to happen (Stadler et al, 2011; Zacks et al, 2011)
IMAGING RESULTS Parametric effects of conditional surprisal Assessing parametric effects of conditional surprisal revealed a positive correlation in the bilateral anterior intraparietal sulcus
Parametric effects of conditional entropy We found a positive correlation of conditional entropy with BOLD response in the right lateral and medial orbitofrontal cortex, dorsomedial prefrontal cortex (dmPFC), bilateral inferior frontal gyrus (IFG), bilateral anterior dorsal insulae, and right posterior intraparietal sulcus; a comprehensive list of activations and Talairach coordinates are provided in Table 2, see Table S5 for MNI coordinates

Summary

Introduction

When we observe another person’s action, we are quite accurate at predicting what is going to happen (Stadler et al, 2011; Zacks et al, 2011). We can acquire knowledge about the structure of action sequences through statistical learning (Avrahami and Kareev, 1994; Baldwin et al, 2008). Thereby, we learn about two statistical measures of actions that we can exploit to predict upcoming steps, given the current action step we observe: the number of possible action steps and their probabilities. After seeing someone grasping a banana, predictability of the step is high, as only one action step is highly probable, while predictability of the action step is lower after seeing someone taking an apple. To keep with the above example, despite the higher predictability after seeing someone grasping a banana, the probability of putting the banana in a lunchbox could be the same as putting an apple in a lunchbox. A differentiation between the two aspects is crucial: while an event’s probability reflects how (un-)expected its occurrence was and how much an observer needs to adapt his previously built expectations, predictability influences how precise an observer’s expectations could be

Objectives

Methods

Results