Journal of Neuroscience最新文献

Attention is key to perception and human behavior, and evidence shows that it periodically samples sensory information (<20 Hz). However, this view has been recently challenged due to methodological concerns and gaps in our understanding of the function and mechanism of rhythmic attention. Here we used an intensive ∼22 h psychophysical protocol combined with reverse correlation analyses to infer the neural representation underlying these rhythms. Participants (male/female) performed a task in which covert spatial (sustained and exploratory) attention was manipulated and then probed at various delays. Our results show that sustained and exploratory attention periodically modulate perception via different neural computations. While sustained attention suppresses distracting stimulus features at the alpha (∼12 Hz) frequency, exploratory attention increases the gain around task-relevant stimulus feature at the theta (∼6 Hz) frequency. These findings reveal that both modes of rhythmic attention differentially shape sensory tuning, expanding the current understanding of the rhythmic sampling theory of attention.

How master splicing regulators cross talk with each other and to what extent transcription regulators are differentially spliced remain unclear in the developing brain. Here, cell-type-specific RNA-Seq analyses of the developing neocortex uncover variable expression of the Rbfox1/2/3 genes and enriched alternative splicing events in transcription regulators, altering protein isoforms or inducing nonsense-mediated mRNA decay. Transient expression of Rbfox proteins in radial glial progenitors induces neuronal splicing events preferentially in transcription regulators such as Meis2 and Tead1 Surprisingly, Rbfox proteins promote the inclusion of a mammal-specific alternative exon and a previously undescribed poison exon in Ptbp1 Simultaneous ablation of Rbfox1/2/3 in the neocortex downregulates neuronal isoforms and disrupts radial neuronal migration. Furthermore, the progenitor isoform of Meis2 promotes Tgfb3 transcription, while the Meis2 neuron isoform promotes neuronal differentiation. These observations indicate that transcription regulators are differentially spliced between cell types in the developing neocortex. (The sex has not been reported to affect cortical neurogenesis in mice, and embryos of both sexes were studied without distinguishing one or the other.).

We aimed to identify the neuronal correlates of performance errors in a difficult timing task. Male rats were trained to seek rewards and avoid shocks depending on the position of photic conditioned stimuli (CS-R and CS-S, respectively). Then, they were exposed to conflict trials where they had to time the interval between the CS-R and CS-S to obtain rewards while avoiding footshocks. There were pronounced individual differences in behavioral strategies on conflict trials. When presented with a CS-S, some rats quickly left the shock sector, forsaking the option of earning a reward, and rarely got shocked. Others earned rewards by delaying avoidance based on the interval between the CS-R and CS-S but were shocked more often. The probability rats would fail a given trial was not stable across trials as rats engaged in incorrect trial runs that were longer than expected by chance. Since this finding suggested that rats shift between two quasi-stable processing modes, we next examined the neuronal correlates of errors. Incorrect trials coincided with reduced firing rates in CA1 and sensory cortical neurons. Moreover, trial-to-trial variations in the firing rates of simultaneously recorded neurons were more strongly correlated on error than correct trials. Last, the power of low-frequency local field potential oscillations was higher during incorrect trials. The finding that the neuronal correlates of correct and error trials are similar in the hippocampus and neocortex lead us to hypothesize that they depend on changes in the activity of common afferents, such as neuromodulatory inputs.

The cerebellum, identified to be active during cognitive and social behavior, has multisynaptic connections through the cerebellar nuclei (CN) and thalamus to cortical regions, yet formation and modulation of these pathways are not fully understood. Perineuronal nets (PNNs) respond to changes in local cellular activity and emerge during development. PNNs are implicated in learning and neurodevelopmental disorders, but their role in the CN during development is unknown. Connectivity deficits, specifically between lateral CN (LCN) and cortical regions have been found in autism spectrum disorder with patients displaying reduced cognitive flexibility. To examine the role of LCN on cognition, neural activity was perturbed in both male and female mice using designer receptors exclusively activated by designer drugs (DREADDs) from postnatal day 21 to 35. We found that while an adolescent LCN disruption did not alter task acquisition, correct choice reversal performance was dependent on DREADD manipulation and sex. Inhibitory DREADDs improved reversal learning in males (5 d faster to criteria), and excitatory DREADDs improved female reversal learning (10 d faster to criteria) compared with controls. Interestingly, the DREADD manipulation in females regardless of direction reduced PNN intensity, whereas in males, only the inhibitory DREADDs reduced PNNs. This suggests a chronic adolescent LCN manipulation may have sex-specific compensatory changes in PNN structure and LCN output to improve reversal learning. This study provides new evidence for LCN in nonmotor functions and sex-dependent differences in behavior and CN plasticity.

To comprehend speech, human brains identify meaningful units, like words, in the speech stream. But whereas the English 'She believed him.' has three words, the Arabic equivalent 'ṣaddaqathu' forms one word with three meaningful subword units, called morphemes: a verb stem ('ṣaddaqa'), a subject suffix ('-t-'), and a direct object pronoun ('-hu'). It remains unclear whether and how speech comprehension involves morpheme processing, above and beyond other language units. Here, we propose and test hierarchically nested encoding models of speech comprehension: a naïve model with word-, syllable-, and sound-level information; a bottom-up model with additional morpheme boundary information; and predictive models that process morphemes before these boundaries. We recorded MEG data as 27 participants (16 female) listened to Arabic sentences like 'ṣaddaqathu .' A temporal response function analysis revealed that in temporal and left inferior frontal regions, predictive models outperform the bottom-up model, which outperforms the naïve model. Moreover, verb stems were either length-ambiguous (e.g., 'ṣaddaqa' is initially mistakable for the shorter stem 'ṣadda', meaning 'blocked') or length-unambiguous (e.g., 'qayyama', meaning 'evaluated', cannot be mistaken for a shorter stem) but shared a uniqueness point, beyond which stem identity is disambiguated. Evoked analyses revealed differences between conditions before the uniqueness point, suggesting that, rather than await disambiguation, the brain employs proactive predictive strategies, processing accumulated input as soon as any possible stem is identifiable, even if not uniquely. These findings highlight the role of morphemes in speech and the importance of including morpheme-level information in neural and computational models of speech comprehension.

Immaturities exist at multiple levels of the developing human visual pathway, starting with immaturities in photon efficiency and spatial sampling in the retina and on through immaturities in the early and later stages of cortical processing. Here, we use steady-state visual evoked potentials (SSVEPs) and controlled visual stimuli to determine the degree to which sensitivity to horizontal retinal disparity is limited by the visibility of the monocular half-images, the ability to encode absolute disparity, or the ability to encode relative disparity. Responses were recorded from male and female human participants at average ages of 5.3 ± 1.6 months, 4.7 ± 1.3 years, and 25.3 ± 6 years. Horizontal disparity sensitivity was measured using planar stereograms that modulated absolute disparity and stereograms portraying disparity gratings that additionally contained relative disparity. Disparity thresholds for absolute disparity changed little over development, but those for relative disparity changed by a factor of ∼10. SSVEPs were also recorded in response to contrast and blur modulation of dynamic random-dot patterns to measure sensitivity to the spatiotemporal content of the monocular half-images. Equating subjective contrast and blur levels between infants, children, and adults based on these measurements did not equate disparity sensitivity. The protracted developmental sequence for horizontal relative disparity coding shown in our measurements is not simply inherited from immaturities in encoding absolute disparity or retinal image contrast but rather reflects immaturities in the computations needed to represent relative disparity that likely involves extrastriate cortical areas where relative disparity is first extracted.

Real-world choice options have many features or attributes, whereas the reward outcome from those options only depends on a few features or attributes. It has been shown that humans learn and combine feature-based with more complex conjunction-based learning to tackle challenges of learning in naturalistic reward environments. However, it remains unclear how different learning strategies interact to determine what features or conjunctions should be attended to and control choice behavior, and how subsequent attentional modulations influence future learning and choice. To address these questions, we examined the behavior of male and female human participants during a three-dimensional learning task in which reward outcomes for different stimuli could be predicted based on a combination of an informative feature and conjunction. Using multiple approaches, we found that both choice behavior and reward probabilities estimated by participants were most accurately described by attention-modulated models that learned the predictive values of both the informative feature and the informative conjunction. Specifically, in the reinforcement learning model that best fit choice data, attention was controlled by the difference in the integrated feature and conjunction values. The resulting attention weights modulated learning by increasing the learning rate on attended features and conjunctions. Critically, modulating decision-making by attention weights did not improve the fit of data, providing little evidence for direct attentional effects on choice. These results suggest that in multidimensional environments, humans direct their attention not only to selectively process reward-predictive attributes but also to find parsimonious representations of the reward contingencies for more efficient learning.

Spiral ganglion neuron (SGN) degeneration is a candidate factor for reduced hearing outcomes in cochlear implant (CI) users. However, there is no procedure available to identify CI contacts close to focal SGN degeneration in human patients. In an animal model, we assessed the impact of focal SGN degeneration on electrical responsiveness and derived an electrophysiological marker for the presence, location, and size of such lesions. We introduced cochlear microlesions in 13 guinea pigs (six female) and recorded electrically evoked compound action potentials (eCAP) after 8-12 d. These were compared with recordings from controls (N = 8) and acutely lesioned cochleae (N = 12). We stimulated via 6-contact CIs in monopolar configuration with symmetric, biphasic pulses of alternating polarity. We histologically assessed the lesion and its relative position to the CI contacts. The lesions (230-850 µm) significantly elevated thresholds and reduced amplitudes. The effect was found at stimulation distances of 3.5 mm from the lesion. A novel eCAP marker, Failure Index (FI: maximal input/output ratio), was significantly elevated in the presence of degenerated SGN. It indicates the failure to efficiently transduce the stimulation current into neural activation (N1P1 amplitude). The FI showed classification accuracies of 80% and identified contacts closest to the lesion in ∼80% of cases within ±700 µm (∼electrode spacing) from the lesion site and was correlated with the lesion size. Thus, the FI is a clinically relevant, noninvasive marker that is suitable for clinical datasets without a priori knowledge on lesions, when combined with the statistical method of median splitting.