Neural correlate of the construction of sentence meaning.

The neural processes that underlie your ability to read and understand this sentence are unknown. Sentence comprehension occurs very rapidly, and can only be understood at a mechanistic level by discovering the precise sequence of underlying computational and neural events. However, we have no continuous and online neural measure of sentence processing with high spatial and temporal resolution. Here we report just such a measure: intracranial recordings from the surface of the human brain show that neural activity, indexed by γ-power, increases monotonically over the course of a sentence as people read it. This steady increase in activity is absent when people read and remember nonword-lists, despite the higher cognitive demand entailed, ruling out accounts in terms of generic attention, working memory, and cognitive load. Response increases are lower for sentence structure without meaning ("Jabberwocky" sentences) and word meaning without sentence structure (word-lists), showing that this effect is not explained by responses to syntax or word meaning alone. Instead, the full effect is found only for sentences, implicating compositional processes of sentence understanding, a striking and unique feature of human language not shared with animal communication systems. This work opens up new avenues for investigating the sequence of neural events that underlie the construction of linguistic meaning.

High-level language processing regions are not engaged in action observation or imitation.

A set of left frontal, temporal, and parietal brain regions respond robustly during language comprehension and production (e.g., Fedorenko E, Hsieh PJ, Nieto-Castañón A, Whitfield-Gabrieli S, Kanwisher N. J Neurophysiol 104: 1177-1194, 2010; Menenti L, Gierhan SM, Segaert K, Hagoort P. Psychol Sci 22: 1173-1182, 2011). These regions have been further shown to be selective for language relative to other cognitive processes, including arithmetic, aspects of executive function, and music perception (e.g., Fedorenko E, Behr MK, Kanwisher N. Proc Natl Acad Sci USA 108: 16428-16433, 2011; Monti MM, Osherson DN. Brain Res 1428: 33-42, 2012). However, one claim about overlap between language and nonlinguistic cognition remains prominent. In particular, some have argued that language processing shares computational demands with action observation and/or execution (e.g., Rizzolatti G, Arbib MA. Trends Neurosci 21: 188-194, 1998; Koechlin E, Jubault T. Neuron 50: 963-974, 2006; Tettamanti M, Weniger D. Cortex 42: 491-494, 2006). However, the evidence for these claims is indirect, based on observing activation for language and action tasks within the same broad anatomical areas (e.g., on the lateral surface of the left frontal lobe). To test whether language indeed shares machinery with action observation/execution, we examined the responses of language brain regions, defined functionally in each individual participant (Fedorenko E, Hsieh PJ, Nieto-Castañón A, Whitfield-Gabrieli S, Kanwisher N. J Neurophysiol 104: 1177-1194, 2010) to action observation ( experiments 1, 2, and 3a) and action imitation ( experiment 3b). With the exception of the language region in the angular gyrus, all language regions, including those in the inferior frontal gyrus (within "Broca's area"), showed little or no response during action observation/imitation. These results add to the growing body of literature suggesting that high-level language regions are highly selective for language processing (see Fedorenko E, Varley R. Ann NY Acad Sci 1369: 132-153, 2016 for a review). NEW & NOTEWORTHY Many have argued for overlap in the machinery used to interpret language and others' actions, either because action observation was a precursor to linguistic communication or because both require interpreting hierarchically-structured stimuli. However, existing evidence is indirect, relying on group analyses or reverse inference. We examined responses to action observation in language regions defined functionally in individual participants and found no response. Thus language comprehension and action observation recruit distinct circuits in the modern brain.

Probabilistic atlas for the language network based on precision fMRI data from >800 individuals.

Two analytic traditions characterize fMRI language research. One relies on averaging activations across individuals. This approach has limitations: because of inter-individual variability in the locations of language areas, any given voxel/vertex in a common brain space is part of the language network in some individuals but in others, may belong to a distinct network. An alternative approach relies on identifying language areas in each individual using a functional 'localizer'. Because of its greater sensitivity, functional resolution, and interpretability, functional localization is gaining popularity, but it is not always feasible, and cannot be applied retroactively to past studies. To bridge these disjoint approaches, we created a probabilistic functional atlas using fMRI data for an extensively validated language localizer in 806 individuals. This atlas enables estimating the probability that any given location in a common space belongs to the language network, and thus can help interpret group-level activation peaks and lesion locations, or select voxels/electrodes for analysis. More meaningful comparisons of findings across studies should increase robustness and replicability in language research.

Toward a universal decoder of linguistic meaning from brain activation.

Prior work decoding linguistic meaning from imaging data has been largely limited to concrete nouns, using similar stimuli for training and testing, from a relatively small number of semantic categories. Here we present a new approach for building a brain decoding system in which words and sentences are represented as vectors in a semantic space constructed from massive text corpora. By efficiently sampling this space to select training stimuli shown to subjects, we maximize the ability to generalize to new meanings from limited imaging data. To validate this approach, we train the system on imaging data of individual concepts, and show it can decode semantic vector representations from imaging data of sentences about a wide variety of both concrete and abstract topics from two separate datasets. These decoded representations are sufficiently detailed to distinguish even semantically similar sentences, and to capture the similarity structure of meaning relationships between sentences.