Uncertainty and expectancy deviations require cortico-subcortical cooperation

Abstract

a b s t r a c tIn a dynamic and uncertain environment it is beneficial to learn the causal structure of the environment in order to minimize uncertainty. This requires determining estimates of probable outcomes, which will guide expectations about incoming information. One key factor in this learning process is to detect whether an unexpected event constitutes a low probability, but valid outcome, or an outright error. The present 7T-fMRI study investigated the role of subcortical structures in regulating this probabilistic inferential learning process. A new task was designed, in which participants learned to calculate the value, and therefore to anticipate the outcome of different visual sequences. Three types of sequences provided unambiguous, ambiguous, and incongruent contextual evidence and each sequence had two outcomes, which differed in their probability of occurrence. We hypothesized that subcortical regions are necessary when expectations are violated, and that their involvement will depend on the nature of the unexpected event. The results show increased dorsomedial striatal and thalamic activation for less probable sequences; in addition, ambiguous sequences also display larger activation in the red nuclei. Incongruent sequences displayed a pattern of subcortical activation restricted to the dorsolateral and the posterior dorsomedial striatum. These results confirm that different subcortical structures regulate uncertainty and expectancy deviations; this is crucial not only for learning to predict events in the environment, but also for flexible cognitive control in general.

Violations of Core Knowledge Shape Early Learning

“…In contrast, when a novel stimulus is paired with an event already fully predicted based on previously experienced contingencies, no error is generated and no learning about the new stimulus occurs (e.g., Kamin, ; Pearce & Hall, ; Rescorla & Wagner, ; Schultz & Dickinson, ). Researchers have identified the neural underpinnings of such error‐driven learning in both non‐human animals (e.g., Friston, ; Hayden, Heilbronner, Pearson, & Platt, ; Holland & Gallagher, ; Hollerman & Schultz, ; Schultz, Dayan, & Montague, ) and human adults, who also generate prediction errors to violations of learned contingencies (e.g., Behrens, Woolrich, Walton, & Rushworth, ; Brown & Braver, ; Carter et al., ; Fletcher et al., ; McClure, Berns, & Montague, ; Mestres‐Missé, Trampel, Turner, & Kotz, ; O'Reilly, ; den Ouden, Friston, Daw, McIntosh, & Stephan, ; Pessiglione, Seymour, Flandin, Dolan, & Frith, ; Yu & Dayan, ). Evidence suggests that in both non‐humans and humans, dopamine neurons play a key role in signaling prediction error and in driving neuronal and behavioral learning (e.g., Holroyd & Coles, ; Waelti, Dickinson, & Schultz, ).…”

Addressing challenges of high spatial resolution UHF fMRI for group analysis of higher‐order cognitive tasks: An inter‐sensory task directing attention between visual and somatosensory domains

“…Spatial smoothing of fMRI data is widely adopted at lower field strength to blur inter-subject structural differences in brain anatomy for group analyses, increase statistical power (Turner & Geyer, 2014), and ensure data meets Gaussian Random Field theory assumptions for statistical analysis (Worsley & Friston, 1995). Until recently, many UHF whole brain studies have employed considerable spatial smoothing (Boyacioglu et al, 2014;Goodman et al, 2017;Mestres-Misse et al, 2017). In a recent study, (Torrisi et al, 2018) showed there is a substantial benefit for smoothing the data at 7 T using the same smoothing kernel at 3 T.…”

“…The challenges of full FOV acquisitions at UHF have resulted in there being a limited number of studies of whole brain function at high spatial resolution to date (e.g., Boyacioglu et al, ; Goodman et al, ; Mestres‐Misse, Trampel, Turner, & Kotz, ; Vu et al, ). In particular there are few studies of cognitive function, as highlighted in a recent review article (De Martino et al, ).…”

Addressing challenges of high spatial resolution, UHF field fMRI for group analysis of higher-order cognitive tasks; an inter-sensory task directing attention between visual and somatosensory domains

“…Spatial smoothing of fMRI data is widely adopted at lower field strength to blur intersubject structural differences in brain anatomy for group analyses, increase statistical power (Turner and Geyer 2014), and ensure data meets Gaussian Random Field theory assumptions for statistical analysis (Worsley and Friston 1995). Until recently, many UHF whole brain studies have employed considerable spatial smoothing (Boyacioglu, Schulz et al 2014, Goodman, Wang et al 2017, Mestres-Misse, Trampel et al 2017).…”

“…Only a limited number of publications have studied whole brain functional responses at UHF (e.g. (Boyacioglu, Schulz et al 2014, Vu, Phillips et al 2016, Goodman, Wang et al 2017, Mestres-Misse, Trampel et al 2017)), with the study of cognitive function being limited, as highlighted in a recent review article (De . To date, to our knowledge, only two studies have used UHF fMRI to map responses to complex cognitive tasks in higher-order cortical regions over the whole brain (Vu, Phillips et al 2016, Goodman, Wang et al 2017).…”