eNeuro最新文献

A central approach in neuroscience is to analyze neural representations as a means to understand a system's function, through the use of methods like principal component analysis, regression, and representational similarity analysis. These analyses often rest on a tacit "linking assumption": that the features explaining the most variance in neural activity are the most important for the system's computation. Here, we challenge this assumption. We review recent work in machine learning demonstrating "representation biases"-the fact that learned representations can be biased toward certain features over others. For example, learned representations heavily overrepresent simple (linear) features while representing complex (nonlinear) features much more weakly, even when both are equally critical for the system's computations. We review the origins of these biases in learning dynamics and patterns of computation. We then discuss their consequences for neuroscience. We show that if a subset of features dominates the representations, standard analytic techniques can yield highly biased inferences-for example, resulting in the mistaken conclusion that a system is simpler than it really is or that two systems are more similar than they really are. We discuss some connections between these findings and recent empirical developments in neuroscience. Finally, we present homomorphic encryption as a conceptual case study of the potential for a total dissociation between representational geometry and computation. We conclude that achieving a complete understanding of neural systems requires moving beyond high-variance signals, as critical computational mechanisms may be hidden in low-variance components.

Consumption varies across the stages (metestrus, diestrus, proestrus, estrus) of a rat's estrous cycle, changing in ways that might be expected to reflect, in part, a direct impact of hormones on taste palatability. Evidence regarding this hypothesis has been mixed, however, and critical within-subject experiments comparing consumption of multiple tastes with distinct valences across all estrous phases have been few. Here, we assayed female rats' licking of palatable (saccharin, sucrose, NaCl) and aversive (quinine-HCl, citric acid) tastes in brief-access trials, while tracking their estrous cycles through vaginal cytology. We observed sucrose palatability to be high at metestrus, the same phase at which the palatability of the aversive citric acid was low. These patterns were consistent across tastes of similar palatability, despite vast differences between the substances' receptor mechanisms and central impacts. Together, these results reveal a general (i.e., independent of particular tastant identity) magnification of palatability-higher than average for palatable tastes and lower for aversive tastes-centered largely on metestrus. We tested whether this phenomenon reflects lateral hypothalamic (LH) estradiol processing, by infusing LH with an estrogen receptor blocker (ICI182, 780) across five consecutive tasting sessions. Control infusions replicated the metestrus magnification of palatability; as predicted, ICI infusions blocked this effect, but estrogen receptor inhibition failed to render preferences "cycle free," instead delaying the palatability magnification until diestrus. In summary, the estrous cycle directly mediates taste palatability in a manner involving hypothalamic actions of estradiol, but this effect is only one of several impacting consumption across the estrous cycle.

Forming a long-term memory requires changes in neuronal transcription. What happens, though, as the memory is forgotten? And how does the transcriptional state relate to the maintenance and recall of the long-term memory? To answer these questions we have been systematically tracing the time-course of transcriptional changes evoked by long-term sensitization in the marine mollusk Aplysia californica Our approach captures transcriptional changes in neurons of known behavioral relevance using a within-subjects design, delineating patterns of transcriptional change that are comprehensive and reproducible. We have previously reported that within 1 day of long-term sensitization training there is a widespread transcriptional response involving robust changes in over 5% of tested transcripts (1,252 of ∼22k; Conte, 2017). Within 1 week, however, memory strength fades and nearly all transcriptional changes relapse to baseline (Perez, 2018). Here we report microarray analysis (N = 16) of transcriptional changes 5 days post-learning, a time-point when memory strength has weakened but is still robust. Remarkably, we find that at this intermediate behavioral stage nearly all transcriptional changes have fully decayed, even in subsets of animals that have shown very little forgetting. Thus, most transcriptional changes seem to decay more rapidly than memory expression. We discuss several possible ways that memory expression could become decoupled from detectable transcriptional regulation.Significance Statement This project characterizes the transcriptional state accompanying a partially-forgotten long-term memory in Aplysia, showing that most transcriptional changes induced during learning fade before forgetting is complete. These results raise interesting questions about the interrelationships between transcriptional, neuronal, and behavioral change.

The suprachiasmatic nucleus (SCN) produces diffusible signals sufficient to sustain circadian locomotor rhythms, though the nature of such signals, their targets, and the pathway whereby such signals may travel is unknown. It is possible that the venous portal veins that connect the capillary beds of the SCN to those of the organum vasculosum of the lamina terminalis (OVLT) provide a vascular pathway whereby signals originating in SCN neurons can reach local targets in the OVLT. Given the presence of the blood-brain interface (BBI) within the SCN, it is unclear how diffusible signals originating in SCN neurons might access the capillary vasculature of this nucleus. Estimates of astrocyte coverage of capillary vasculature range widely, from 70-100%, and furthermore such coverage can change dynamically. In the present study, we investigated whether three vasoactive peptidergic processes found in the mouse SCN, namely vasopressin, vasoactive intestinal peptide and gastrin releasing peptide, might breach the BBI thereby accessing capillary vessels. Using widefield and confocal imaging, we found neuron-to-capillary contacts between varicosities bearing each of these vasoactive peptides and capillary basal membranes, pericytes and the endothelia in the mouse SCN of either sex. The findings suggest that all three vasoactive peptides may functionally breach the BBI of the SCN highlighting the importance of understanding how these peptides act on local vasculature to impact blood flow.Significance Statement The suprachiasmatic nucleus (SCN) produces diffusible signals sufficient to sustain circadian locomotor rhythms. The SCN - organum vasculosum lamina terminalis portal pathway provides a route whereby signals of SCN origin might be relayed to the rest of the brain. The presence of the blood-brain interface (BBI), however, raises the question of how diffusible signals of SCN origin might access the capillary vasculature of the nucleus. High resolution confocal imaging results suggest that varicosities of vasoactive peptides found in the SCN, namely vasopressin, vasoactive intestinal peptide and gastrin releasing peptide, breach the BBI and directly contact capillary vessels compartments including basal laminae, pericytes and endothelia.

A possible mechanism of temporal lobe epilepsy is insufficient inhibition of hippocampal dentate granule cells. Precipitating injuries that kill interneurons in the dentate gyrus might result in fewer inhibitory synapses with granule cells. To test this hypothesis, previous studies evaluated numbers or densities of interneurons, γ-amino butyric acid (GABA)ergic boutons, and inhibitory synapses in tissue from human patients with temporal lobe epilepsy and rodent models. However, those studies have limitations. Some of those limitations can be addressed by a large animal model. Sea lions (Zalophus californianus) can develop temporal lobe epilepsy naturally. Like humans, epileptic sea lions exhibit bilateral or unilateral hippocampal sclerosis (neuron loss) with granule cell vulnerability, but sea lions permit optimal tissue preservation and sampling, and good control subjects. To label interneuron cell bodies and GABAergic synaptic boutons, sea lion hippocampal tissue from both sexes was processed with immunohistochemistry for glutamic acid decarboxylase (GAD) and vesicular GABA transporter. Stereological techniques were used to evaluate the dentate gyrus of the entire hippocampus. Numbers of granule cells, GAD cells, and GABAergic boutons were substantially reduced in shrunken, sclerotic hippocampi. However, numbers of GABAergic boutons and granule cells were correlated. These findings indicate that, despite losses, numbers of GABAergic boutons scale with numbers of surviving granule cells.Significance Statement Temporal lobe epilepsy is a challenging clinical problem. Electrophysiological studies reveal that hippocampal dentate granule cells are insufficiently inhibited and hyperexcitable in epileptic tissue from humans and rodent models. The present stereological analysis of a large animal model (sea lions) found no evidence for disproportionate loss of dentate gyrus GABAergic boutons in temporal lobe epilepsy. These data suggest reduced inhibition of granule cells is attributable to something other than too few GABAergic boutons.

Repeated restraint stress (RRS) in rats impairs cognitive flexibility, particularly when faced with additional mild acute stress (AS). We tested the hypothesis that this impairment is associated with altered dopaminergic activity in the dorsal striatum (DS) driven by corticotropin-releasing-factor receptor type 1 (CRFR1) in the substantia nigra pars compacta (SNpc). Sixty-two male rats received RRS or handling for 14 d, before training on a two-action, two-outcome instrumental conditioning task. Initial learning was assessed using an outcome devaluation test. Cognitive flexibility was assessed by reversing the outcome identities and a second outcome devaluation test, with half the rats in each group receiving AS before reversal training. Dopamine and metabolites were quantified in the DS, and CRFR1 mRNA was quantified in the SNpc. In Experiment 2, SNpc CRFR1 was pharmacologically blocked unilaterally before AS and reversal training in 32 male and 32 female rats. Increased dopaminergic activity in the DS and SNpc and CRFR1 expression in the SNpc in the left hemisphere were associated with resilience to AS in naive rats, but with an impairment in RRS + AS rats. Blocking CRFR1 in the left SNpc impaired cognitive flexibility following AS in naive rats but restored it following AS in RRS rats. Blocking CRFR1 in the SNpc increased DA availability in the DMS but decreased it in the DLS. The study suggests opposite facilitation in DA availability in the medial and lateral DS by CRFR1 in the SNpc and a left-to-right transition in dopaminergic nigrostriatal projection activity as a protective mechanism following RRS.

Reward prediction errors (RPEs) guide learning by comparing expected and obtained outcomes. In mammals, ventral tegmental area (VTA) activity is closely linked to RPE-like signaling, yet how avian VTA dynamics evolve during reinforcement learning remains less well characterized. Here we recorded VTA spiking in pigeons (two females and one male) performing a cue-guided operant task in which a green cue (cue⁺) predicted reward contingent on a key peck, whereas a red cue (cue^-) was unrewarded. Using a 16-channel microwire array, we analyzed pooled channel-level multiunit activity (MUA) aligned to task events. Across sessions, cue⁺ trials showed a learning-related redistribution of event-locked modulation: outcome-locked activity was more prominent early in training, while cue-locked modulation became stronger as performance stabilized, consistent with a temporal-difference-like shift of prediction-related signals. Cue^- trials were sparse after early learning and showed limited cue-locked modulation in the available dataset. Together, these results provide initial evidence that pigeon VTA pooled MUA exhibits learning-related dynamics consistent with RPE-like processing and support cross-species comparisons of dopaminergic learning signals.

Presbycusis, a prevalent neurodegenerative disorder, is characterized by declining speech recognition and has been associated with cognitive impairments across multiple domains. However, the underlying neurobiological mechanisms between presbycusis and cognitive impairments remain unclear. We assessed pure-tone audiometry thresholds (PTA), speech recognition thresholds (SRT), and cognitive abilities in individuals with presbycusis (24 males and 31 females) and healthy controls (23 males and 32 females). Using magnetic resonance imaging, we calculated the amplitude of low-frequency fluctuations (ALFF) to characterize function and gray matter volume (GMV) to characterize structure. Based on ALFF and GMV, we calculated functional-structural ratio (FSR) to measure the functional-structural coupling. Significant correlations between GMV atrophy and ALFF changed in the putamen, fusiform gyrus, precuneus, and medial superior frontal gyrus in presbycusis group, and these changes were significantly associated with the increase in PTA and SRT. The FSR reduction in the FFG, precuneus, and medial superior frontal gyrus were also significantly associated with the increase in PTA and SRT. Moreover, it was also significantly correlated with lower scores on the Montreal Cognitive Assessment (MoCA) and the Auditory Verbal Learning Test (AVLT), as well as the prolonged time in the Trail Making Test (TMT-A). Presbycusis involves coupled structural atrophy and functional decline in auditory and higher-order cognitive regions. Crucially, reduced FSR correlates with both worsening hearing thresholds and cognitive impairment. This highlights FSR as a key neurobiological link between hearing loss and cognitive decline. This research provides a novel basis for early screening and dynamic monitoring of presbycusis-related cognitive impairment.

Magnetic resonance imaging (MRI) is a critical tool for translational neuroscience, but preclinical studies frequently rely on anesthesia, which alters neural activity and limits comparison with human studies. Awake rodent functional MRI (fMRI) enables investigation of brain function under physiologically relevant conditions; however, its implementation is constrained by the need for anesthesia during restraint setup. We developed and evaluated a restraint system and habituation protocol for awake rat fMRI. Ten rats were studied: an awake group and an anesthetized group (three males and two females per group). The protocol included head post implantation and an 11 d habituation period. T2-weighted anatomical and functional scans were acquired. Head motion and functional connectivity were analyzed using the RABIES pipeline and compared between groups. The modular 3D-printed restraint system developed can be assembled in under 5 min; eliminates the need for anesthesia, ear bars, and bite bars; and supports several behavioral paradigms. High-quality anatomical and functional images were obtained for awake rats. Anesthetized rats exhibited significantly lower translation, rotation, and framewise displacement. Functional connectivity differed between awake and anesthetized rats, with some region pairs showing higher (e.g., left-right primary somatosensory cortex and hypothalamus-insula) and lower (e.g., cingulate-prelimbic cortex and retrosplenial-motor cortex) correlations in awake rats. However, these differences did not survive network-based statistics correction. This work presents a scalable, reproducible, and animal-friendly platform for awake rat fMRI that enables high-quality, behaviorally enriched imaging without anesthesia, while highlighting the effects of anesthesia on functional connectivity.