Symposium: Computational models of MMN
Friday, Sep 11, 2015
14:30-15:30
Hörsaal 3
MMN: a marker of statistical learning in the brain
1Queensland Brain Institute, The University of Queensland, Brisbane, Australia
2Duke-National University of Singapore, Singapore, Singapore
3The University of Queensland, Brisbane, Australia
m.garrido@uq.edu.au
The ability to learn about regularities in the environment is fundamental for adaptive behaviour. Neural responses to unpredictable sensory events carry a unique prediction error signature, as the mismatch negativity (MMN) in classic "oddball" paradigms. In the real world, however, learning about regularities often occurs in the context of competing cognitive demands. Here we asked whether learning of statistical regularities is modulated by concurrent cognitive load. Across two experiments, we compared electroencephalographic (EEG) metrics associated with responses to pure tones with frequencies sampled from narrow or wide Gaussian distributions. In Experiment 1, we replicated our previous finding that tones in the tails of the distributions ("oddballs") evoked a larger response than those in the center ("standards"). Moreover, this prediction error response (MMN) was larger for physically identical outliers in the narrow than in the wide distribution. These results suggest that observers can track the uncertainty associated with distributions of apparently random sensory events. In Experiment 2, participants performed a N-back task on a central letter stream while listening to the same sequences presented in Experiment 1. Here we compared "standards" and "oddballs" under different distributional variance (narrow and wide) and cognitive loads (low and high). We again observed reliable MMN to outliers that were greater when the distribution was narrower, suggesting that observers were able to track statistical uncertainty under cognitive load. Our findings suggest that statistical learning is not a capacity limited process, and that it might proceed automatically even when cognitive resources are taxed by concurrent demands.