Purchase this article with an account.
Ying Joey Zhou, Alexis Pérez-Bellido, Saskia Haegens, Floris P de Lange; No modulation by expectation of the sensory response to object images as measured by MEG. Journal of Vision 2019;19(10):271c. doi: https://doi.org/10.1167/19.10.271c.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Several studies have found that expected stimuli evoke a reduced neural response compared to unexpected stimuli (“expectation suppression”). Using a statistical learning paradigm, we asked whether this neural response difference between expected and unexpected stimuli stems from i) a suppression of the expected, ii) an enhancement of the unexpected or iii) both. On Day 1, participants (N=34) were exposed to 9 leading and 6 trailing images, which were paired such that a leading image was followed either by a specific trailing image (“expected” pair), or by any of the trailing images equally likely (“neutral” pair). Participants’ task was to respond when an image was presented upside-down. On Day 2, participants performed the same task, while their neurophysiological activity was recorded with MEG. The same images used on Day 1 were presented, with the only difference being that the leading images of those “expected” pairs were occasionally (8%) followed by any trailing image not paired with them before (“unexpected” pair). Immediately after MEG recording, participants were tested on their knowledge about the image pairs. Surprisingly, we did not observe any modulation of sensory activity by expectation, in either event-related fields, oscillatory low-frequency activity, or high-frequency band activity (70–150 Hz). Our results are unlikely due to participants’ ignorance of the predictive relationship between images, given that participants exhibited reliable behavioral benefits for expected pairs in the post-MEG test, when they had to categorize the trailing images as fast as possible. These results are surprising given that robust expectation suppression effects in the visual ventral stream have been reported with statistical learning paradigms in both human fMRI and primate single-cell studies. As EEG/MEG activity primarily reflects population-level post-synaptic potentials generated within the apical dendrites of pyramidal neurons, our results may constrain the type and neural locus of neural activity modulations induced by expectation.
This PDF is available to Subscribers Only