Postersession 2
Poster #: 56
Topic: Error signals
Thursday, Sep 10, 2015
1st floor

High gamma frontal cortex activity dissociates predicted vs. unpredicted deviation: an intracranial EEG study

Stefan Dürschmid1, Leon Deouell2, Hermann Hinrichs3, Hans-Jochen Heinz, & Robert T Knight

1Neurology, Otto-von-Guericke University, Magdeburg, Germany
2Department of Psychology and Edmond and Lily Safra Center for brain sciences, The Hebrew University of Jerusalem, Jerusalem, Israel
3Magdeburg, Germany

Predictive coding theories imply that stimuli deviating from expectancy result in a prediction-error (PE) signal. Previous research on processing of auditory deviancy has capitalized on low frequency event-related components (LFs) as signatures of PE. However, local cortical neuronal activity (NA) is best indexed by higher frequency bands (high gamma: Hγ) activity. Utilizing the high temporal and spectral resolution of direct cortical recordings from subdural electrodes (ECoG), we compared the temporal patterns of LFs and Hγ. Subjects (n=5) heard trains of task-irrelevant auditory stimuli in two conditions differing in the predictability of deviation from repetitive background stimuli (fully predictable vs. unpredictable deviants). Subjects were instructed to ignore the sounds and watch a visual slide show. Electrodes with significant Hγ and/or LF increase over baseline were classified as task-active Hγ/LF channels. Principal component analysis (PCA) was used to identify consistent temporal patterns of Hγ/LF activation. Deviant stimuli elicited Hγ activity earlier in time than LFs. Critically, only frontal Hγ activity discriminated between fully predictable and unpredictable changes. Furthermore, Hγ activity occurring 100 msec before fully predictable deviants differed from activity before standards and unpredictable deviants. ROC analysis revealed that middle frontal cortex prestimulus activity predicts fully predictable deviants on a single trial level. In distinction, the inferior frontal cortex was sensitive to the increasing likelihood of deviants following long trains of standards in the unpredictable condition. We propose that prefrontal cortex plays a critical role in prediction of sensory input even when the auditory stimuli are task irrelevant.