Linguistic analyses suggest that sentences are not mere strings of words but possess a hierarchical structure with constituents nested inside each other. We used functional magnetic resonance imaging (fMRI) to search for the cerebral mechanisms of this theoretical construct. We hypothesized that the neural assembly that encodes a constituent grows with its size, which can be approximately indexed by the number of words it encompasses. We therefore searched for brain regions where activation increased parametrically with the size of linguistic constituents, in response to a visual stream always comprising 12 written words or pseudowords. The results isolated a network of left-hemispheric regions that could be dissociated into two major subsets. Inferior frontal and posterior temporal regions showed constituent size effects regardless of whether actual content words were present or were replaced by pseudowords (jabberwocky stimuli). This observation suggests that these areas operate autonomously of other language areas and can extract abstract syntactic frames based on function words and morphological information alone. On the other hand, regions in the temporal pole, anterior superior temporal sulcus and temporo-parietal junction showed constituent size effect only in the presence of lexico-semantic information, suggesting that they may encode semantic constituents. In several inferior frontal and superior temporal regions, activation was delayed in response to the largest constituent structures, suggesting that nested linguistic structures take increasingly longer time to be computed and that these delays can be measured with fMRI.

Making decisions in uncertain environments often requires combining multiple pieces of ambiguous information from external cues. In such conditions, human choices resemble optimal Bayesian inference, but typically show a large suboptimal variability whose origin remains poorly understood. In particular, this choice suboptimality might arise from imperfections in mental inference rather than in peripheral stages, such as sensory processing and response selection. Here, we dissociate these three sources of suboptimality in human choices based on combining multiple ambiguous cues. Using a novel quantitative approach for identifying the origin and structure of choice variability, we show that imperfections in inference alone cause a dominant fraction of suboptimal choices. Furthermore, two-thirds of this suboptimality appear to derive from the limited precision of neural computations implementing inference rather than from systematic deviations from Bayes-optimal inference. These findings set an upper bound on the accuracy and ultimate predictability of human choices in uncertain environments.