Institut du Cerveau

We present a perspective that integrates clinical and neuroimaging evidence to argue that voluntary visual mental imagery is supported by a distributed network organized around the Fusiform Imagery Node (FIN) of the left fusiform gyrus. Lesion evidence shows that imagery is consistently impaired when the FIN is damaged or disconnected, while neuroimaging confirms robust, domain-independent activation of this cortical area. Recent 7T fMRI data further clarify the putative functional attributes of the FIN. First, the FIN appears to subserve domain-general processes, supporting imagery for objects, colors, words, faces, and spatial relationships, whereas adjacent ventral temporal regions display domain selectivity. Second, it encodes semantic information, as shown by shared multivoxel activity patterns between the FIN, the left inferior frontal gyrus, and the intraparietal sulcus, indicating common representational codes for semantic content. Third, FIN representational overlap between imagery and perception correlates with subjective vividness-the more perceptual-like the representation, the more vivid the experience. This overlap is absent in congenital aphantasia. Fourth, the FIN exhibits functional connectivity with both frontoparietal control regions and domain-specific areas, consistent with a role as a semantic integration hub. Notably, FIN-prefrontal connectivity is markedly reduced in aphantasia. Fifth, the FIN is strongly left-lateralized, mirroring the asymmetry of the semantic system that provides a major input to imagery. Collectively, these features suggest that the FIN serves as a central bridge between semantic and visual information, essential for constructing mental images. Its connectivity may account for individual variability in imagery vividness-from aphantasia to typical imagery-and offers theoretical insight into the neural mechanisms of conscious processing and hemispheric asymmetries, with implications for both clinical and applied domains.

Abstract Creativity is a hallmark of human cognition, characterized by the ability to connect seemingly distant concepts or ideas. Existing theories suggest that remote thinking can be achieved either spontaneously (constrained by the structure of semantic memory) or in a goal-directed manner (constrained by a creative goal). The present study investigates the neural correlates of goal-directed remote thinking, defined as the intentional production of semantically distant associations. Using a simple word-to-word association task comprising both a spontaneous condition and a goal-directed creative condition, we investigated Goal-directed Remoteness as the extra semantic distance traveled away from spontaneous responses when instructed to think creatively. Task-based functional MRI in 38 healthy young adults identified brain regions whose activation scaled with Goal-directed Remoteness . The results revealed that activity in the rostromedial and dorsomedial prefrontal cortex and the right cerebellar Crus I & II was positively modulated by Goal-directed Remoteness. Control analyses confirmed the robustness of these findings independently of the cue-words’ semantic or linguistic properties. Follow-up seed-based resting-state functional connectivity analyses characterized the intrinsic connectivity profiles of the revealed regions. They showed that rostromedial and dorsomedial prefrontal and cerebellar Crus I & II regions formed a functionally interconnected network primarily overlapping with the default-mode network (DMN). Our findings challenge the traditional view of the DMN as supporting only passive or spontaneous cognition. Instead, they reveal a prefronto-cerebellar DMN subnetwork supporting goal-directed remote thinking, a key component of creative cognition. Within this network, the rostromedial and dorsomedial prefrontal cortex and cerebellar Crus I & II play an active role in the intentional generation of connections between distant concepts.

The germline-restricted chromosome (GRC) of passerines is a remarkable tissue-specific chromosome that accumulated paralogs of genes from the regular "A chromosomes" over millions of years, often amplified into dozens of gene copies. In addition to its repetitive content, typically uniparental inheritance, and lack of recombination, the GRC resembles non-recombining sex chromosomes and some B chromosomes, for all of which assembly and single-nucleotide polymorphisms (SNPs) calling are difficult. Here, we first show that much of the Australian zebra finch macro-GRC can be assembled using accurate long reads. We then describe a paralog-aware Snakemake pipeline, ParaVar, to map short reads from the GRC to retrieve GRC regions suitable for haplotype-based analysis. ParaVar reliably calls hundreds of SNPs across the GRC, thereby providing an estimate of nucleotide diversity on the highly repetitive zebra finch macro-GRC. Our results show significantly lower nucleotide diversity (20- to 50-fold lower) on the GRC compared to the mitogenome and autosomes, and a strong phylogenetic discordance between the GRC and the mitochondrial genome. Beyond the contribution of background selection, our results suggest that a single GRC haplotype recently spread through the populations while jumping across matrilines via occasional paternal inheritance. We anticipate that our paralog-aware pipeline will be useful for SNP calling and population genetics analyses of repetitive GRCs, sex chromosomes, and B chromosomes.

NCT03811769.