Learning brain dynamics for decoding and predicting individual differences

Insights from functional Magnetic Resonance Imaging (fMRI), as well as recordings of large numbers of neurons, reveal that many cognitive, emotional, and motor functions depend on the multivariate interactions of brain signals. To decode brain dynamics, we propose an architecture based on recurrent neural networks to uncover distributed spatiotemporal signatures. We demonstrate the potential of the approach using human fMRI data during movie-watching data and a continuous experimental paradigm. The model was able to learn spatiotemporal patterns that supported 15-way movie-clip classification (∼90%) at the level of brain regions, and binary classification of experimental conditions (∼60%) at the level of voxels. The model was also able to learn individual differences in measures of fluid intelligence and verbal IQ at levels comparable to that of existing techniques. We propose a dimensionality reduction approach that uncovers low-dimensional trajectories and captures essential informational (i.e., classification related) properties of brain dynamics. Finally, saliency maps and lesion analysis were employed to characterize brain-region/voxel importance, and uncovered how dynamic but consistent changes in fMRI activation influenced decoding performance. When applied at the level of voxels, our framework implements a dynamic version of multivariate pattern analysis. Our approach provides a framework for visualizing, analyzing, and discovering dynamic spatially distributed brain representations during naturalistic conditions.     

Individual differences in estrogen receptor alpha in select brain nuclei are associated with individual differences in aggression

Steroid hormones play an important role in modulating social behavior in many species. Estrogens are thought to act on an interconnected network of hypothalamic and limbic brain areas to affect aggressive behavior, although the specific nuclei unknown remain unspecified. We show that individual variation in estrogen receptor alpha (ERalpha) immunoreactivity in the lateral septum (LS), ventral bed nucleus of the stria terminalis (vBNST), and anterior hypothalamus (AHA) of CD-1 mice is positively correlated with aggressive behavior. When males were treated with fadrozole (an aromatase inhibitor), aggressive behavior was reduced, although castration did not reduce aggression. These results suggest that estrogens modulate aggressive behavior by acting on a circuit that includes the LS, vBNST, and AHA and that the source of estrogens is non-gonadal. Fadrozole also decreased c-fos expression in the lateral septum following aggressive encounters. Although the effects of estrogen on aggression appear to involve regulation of neuronal activity in the LS, additional processes are likely involved. These results suggest that estrogen acts in a specific subset of a complex network of nuclei to affect aggressive behavior.     

Sex differences and autism: brain function during verbal fluency and mental rotation

Autism spectrum conditions (ASC) affect more males than females. This suggests that the neurobiology of autism: 1) may overlap with mechanisms underlying typical sex-differentiation or 2) alternately reflect sex-specificity in how autism is expressed in males and females. Here we used functional magnetic resonance imaging (fMRI) to test these alternate hypotheses. Fifteen men and fourteen women with Asperger syndrome (AS), and sixteen typically developing men and sixteen typically developing women underwent fMRI during performance of mental rotation and verbal fluency tasks. All groups performed the tasks equally well. On the verbal fluency task, despite equivalent task-performance, both males and females with AS showed enhanced activation of left occipitoparietal and inferior prefrontal activity compared to controls. During mental rotation, there was a significant diagnosis-by-sex interaction across occipital, temporal, parietal, middle frontal regions, with greater activation in AS males and typical females compared to AS females and typical males. These findings suggest a complex relationship between autism and sex that is differentially expressed in verbal and visuospatial domains.     

Purine metabolism abnormalities in a hyperuricosuric subclass of autism

A subclass of patients with classic infantile autism have uric acid excretion which is >2 S.D.s above the normal mean. These hyperuricosuric autistic individuals may comprise approx. 20% of the autistic population. In order to determine the metabolic basis for urate overexcretion in these patients, de novo purine synthesis was measured in the cultured skin fibroblasts of these patients by quantification of the radiolabeled purine compounds produced by incubation with radiolabeled sodium formate. For comparison, de novo purine synthesis in normal controls, in normouricosuric autistic patients, and cells from patients with other disorders in which excessive uric acid excretion is seen was also measured. These experiments showed that de novo purine synthesis is increased approx. 4-fold in the hyperuricosuric autistic patients. This increase was less than that found in other hyperuricosuric disorders. No unusual radiolabeled compounds (such as adenylosuccinate) were detected in these experiments, and no gross deficiencies of radiolabeled nucleotides were seen. However, the ratio of adenine to guanine nucleotides produced by de novo synthesis was found to be lower in the cells of the hyperuricosuric autistic patients than in the normal controls or the cells from patients with other disorders. These results indicate that the hyperuricosuric subclass of autistic patients have increased de novo purine synthesis, and that the increase is approximately that expected for the degree of urate overexcretion when compared to other hyperuricosuric disorders. No particular enzyme defect was suggested by either gross deficiency of a radiolabeled compound or the appearance of an unusual radiolabeled compound, and no potentially neurotoxic metabolites were seen. Although an enzyme defect responsible for the accelerated purine synthesis was not identified, the abnormal ratio of adenine to guanine nucleotides suggests a defect in purine nucleotide interconversion.     

Mapping early brain development in autism

Although the neurobiology of autism has been studied for more than two decades, the majority of these studies have examined brain structure 10, 20, or more years after the onset of clinical symptoms. The pathological biology that causes autism remains unknown, but its signature is likely to be most evident during the first years of life when clinical symptoms are emerging. This review highlights neurobiological findings during the first years of life and emphasizes early brain overgrowth as a key factor in the pathobiology of autism. We speculate that excess neuron numbers may be one possible cause of early brain overgrowth and produce defects in neural patterning and wiring, with exuberant local and short-distance cortical interactions impeding the function of large-scale, long-distance interactions between brain regions. Because large-scale networks underlie socio-emotional and communication functions, such alterations in brain architecture could relate to the early clinical manifestations of autism. As such, autism may additionally provide unique insight into genetic and developmental processes that shape early neural wiring patterns and make possible higher-order social, emotional, and communication functions.     

Functional-anatomic correlates of individual differences in memory

Memory abilities differ greatly across individuals. To explore a source of these differences, we characterized the varied strategies people adopt during unconstrained encoding. Participants intentionally encoded object pairs during functional MRI. Principal components analysis applied to a strategy questionnaire revealed that participants variably used four main strategies to aid learning. Individuals' use of verbal elaboration and visual inspection strategies independently correlated with their memory performance. Verbal elaboration correlated with activity in a network of regions that included prefrontal regions associated with controlled verbal processing, while visual inspection correlated with activity in a network of regions that included an extrastriate region associated with object processing. Activity in regions associated with use of these strategies was also correlated with memory performance. This study reveals functional-anatomic correlates of verbal and perceptual strategies that are variably used by individuals during encoding. These strategies engage distinct brain regions and may separately influence memory performance.     

Note: individual differences in taste adaptation

Subjects who had previously participated in a taste adaptationstudy (DuBose, et al., 1977) were retested one year later usingthe same stimuli (.1 M and .36 M sucrose and NaCl) and experimentalconditions (3-minute continuous flow over anterior dorsal tonguesurface). Results indicated that individual differences in thereported degree of adaptation were maintained over the longintersession interval. Salivary sodium levels and salt recognitionthresholds could not account for the persisting individual differencesin adaptation to NaCl. Direct examination of subjects' tonguemovements is suggested.     

Genetics and mitochondrial abnormalities in autism spectrum disorders: a review

We review the current status of the role and function of the mitochondrial DNA (mtDNA) in the etiology of autism spectrum disorders (ASD) and the interaction of nuclear and mitochondrial genes. High lactate levels reported in about one in five children with ASD may indicate involvement of the mitochondria in energy metabolism and brain development. Mitochondrial disturbances include depletion, decreased quantity or mutations of mtDNA producing defects in biochemical reactions within the mitochondria. A subset of individuals with ASD manifests copy number variation or small DNA deletions/duplications, but fewer than 20 percent are diagnosed with a single gene condition such as fragile X syndrome. The remaining individuals with ASD have chromosomal abnormalities (e.g., 15q11-q13 duplications), other genetic or multigenic causes or epigenetic defects. Next generation DNA sequencing techniques will enable better characterization of genetic and molecular anomalies in ASD, including defects in the mitochondrial genome particularly in younger children.     

EEG abnormalities, epilepsy and regression in autism: a review

Autism is associated with a high frequency of epileptiform EEG abnormalities (prevalence range 10.3-72.4%) and epilepsy (prevalence range 0-44.5%). A significant subgroup of autistic children (20-49%) experience autistic regression. The relationship among EEG abnormalities, epilepsy, and regression in autistic patients is not yet well understood. In this review, the current knowledge of the relationship is summarized. The evidence from clinical studies does not support the view that EEG abnormalities play a role in autistic regression. The majority of studies also failed to find any significant relationship between epilepsy and autistic regression. However, some results indicated that the higher the prevalence of epilepsy in the sample, the greater the probability of there being a significant association between epilepsy and autistic regression. Further research on the topic is needed.     

Adaptive individual differences within single populations

Phenotypic differences can exist between species, between local populations of the same species and between individuals within single local populations. At all scales, phenotypic differences can be either adaptive or non-adaptive. Using natural selection to explain differences between closely related species was controversial during the 1940s but had become common by the 1960s. Similarly, the adaptive nature of differences between local populations was initially controversial but had become widely accepted by the 1980s. The interpretation of differences at the finest scale, between individuals within single populations, is still unresolved. This paper reviews studies of adaptive individual differences in resource use and response to risk. A general conceptual framework for thinking about adaptive individual differences within populations can unite subjects as seemingly different as speciation and personality psychology.     

An ontogenetic perspective on individual differences

Phenotypic differences among individuals can arise during any stage of life. Although several distinct processes underlying individual differences have been defined and studied (e.g. parental effects, senescence), we lack an explicit, unified perspective for understanding how these processes contribute separately and synergistically to observed variation in functional traits. We propose a conceptual framework based on a developmental view of life-history variation, linking each ontogenetic stage with the types of individual differences originating during that period. In our view, the salient differences among these types are encapsulated by three key criteria: timing of onset, when fitness consequences are realized, and potential for reversibility. To fill a critical gap in this framework, we formulate a new term to refer to individual differences generated during adulthood—reversible state effects. We define these as ‘reversible changes in a functional trait resulting from life-history trade-offs during adulthood that affect fitness’, highlighting how the adult phenotype can be repeatedly altered in response to environmental variation. Defining individual differences in terms of trade-offs allows explicit predictions regarding when and where fitness consequences should be expected. Moreover, viewing individual differences in a developmental context highlights how different processes can work in concert to shape phenotype and fitness, and lays a foundation for research linking individual differences to ecological and evolutionary theory.     

An evolutionary ecology of individual differences

Individuals often differ in what they do. This has been recognised since antiquity. Nevertheless, the ecological and evolutionary significance of such variation is attracting widespread interest, which is burgeoning to an extent that is fragmenting the literature. As a first attempt at synthesis, we focus on individual differences in behaviour within populations that exceed the day-to-day variation in individual behaviour (i.e. behavioural specialisation). Indeed, the factors promoting ecologically relevant behavioural specialisation within natural populations are likely to have far-reaching ecological and evolutionary consequences. We discuss such individual differences from three distinct perspectives: individual niche specialisations, the division of labour within insect societies and animal personality variation. In the process, while recognising that each area has its own unique motivations, we identify a number of opportunities for productive 'cross-fertilisation' among the (largely independent) bodies of work. We conclude that a complete understanding of evolutionarily and ecologically relevant individual differences must specify how ecological interactions impact the basic biological process (e.g. Darwinian selection, development and information processing) that underpin the organismal features determining behavioural specialisations. Moreover, there is likely to be co-variation amongst behavioural specialisations. Thus, we sketch the key elements of a general framework for studying the evolutionary ecology of individual differences.     

Psychoactive pollution suppresses individual differences in fish behaviour

Environmental contamination by pharmaceuticals is global, substantially altering crucial behaviours in animals and impacting on their reproduction and survival. A key question is whether the consequences of these pollutants extend beyond mean behavioural changes, restraining differences in behaviour between individuals. In a controlled, two-year, multigenerational experiment with independent mesocosm populations, we exposed guppies (Poecilia reticulata) to environmentally realistic levels of the ubiquitous pollutant fluoxetine (Prozac). Fish (unexposed: n = 59, low fluoxetine: n = 57, high fluoxetine: n = 58) were repeatedly assayed on four separate occasions for activity and risk-taking behaviour. Fluoxetine homogenized individuals'' activity, with individual variation in populations exposed to even low concentrations falling to less than half that in unexposed populations. To understand the proximate mechanism underlying these changes, we tested the relative contribution of variation within and between individuals to the overall decline in individual variation. We found strong evidence that fluoxetine erodes variation in activity between but not within individuals, revealing the hidden consequences of a ubiquitous contaminant on phenotypic variation in fish—likely to impair adaptive potential to environmental change.     

The importance of individual differences in grapheme-color synesthesia

In this issue of Neuron, Hubbard et al. show individual differences in how grapheme-color synesthetes perform on cognitive tasks. Importantly, these behavioral differences were correlated with fMRI measures. Such individual differences have important ramifications for synesthesia research. If individual differences are ignored, then synesthesia research will be characterized by erroneous conclusions and failures to replicate.     

Mind blindness and the brain in autism.

Experimental evidence shows that the inability to attribute mental states, such as desires and beliefs, to self and others (mentalizing) explains the social and communication impairments of individuals with autism. Brain imaging studies in normal volunteers highlight a circumscribed network that is active during mentalizing and links medial prefrontal regions with posterior superior temporal sulcus and temporal poles. The brain abnormality that results in mentalizing failure in autism may involve weak connections between components of this system.