A neuro-inspired computational model of life-long learning and catastrophic interference, mimicking hippocampus novelty-based dopamine modulation and lateral inhibitory plasticity.

The human brain has a remarkable lifelong learning capability to acquire new experiences while retaining previously acquired information. Several hypotheses have been proposed to explain this capability, but the underlying mechanisms are still unclear. Here, we propose a neuro-inspired firing-rate computational model involving the hippocampus and surrounding areas, that encompasses two key mechanisms possibly underlying this capability. The first is based on signals encoded by the neuromodulator dopamine, which is released by novel stimuli and enhances plasticity only when needed. The second is based on a homeostatic plasticity mechanism that involves the lateral inhibitory connections of the pyramidal neurons of the hippocampus. These mechanisms tend to protect neurons that have already been heavily employed in encoding previous experiences. The model was tested with images from the MNIST machine learning dataset, and with more naturalistic images, for its ability to mitigate catastrophic interference in lifelong learning. The results show that the proposed biologically grounded mechanisms can effectively enhance the learning of new stimuli while protecting previously acquired knowledge. The proposed mechanisms could be investigated in future empirical animal experiments and inspire machine learning models.

A Biologically Inspired Neural Network Model to Gain Insight Into the Mechanisms of Post-Traumatic Stress Disorder and Eye Movement Desensitization and Reprocessing Therapy.

Eye movement desensitization and reprocessing (EMDR) therapy is a well-established therapeutic method to treat post-traumatic stress disorder (PTSD). However, how EMDR exerts its therapeutic action has been studied in many types of research but still needs to be completely understood. This is in part due to limited knowledge of the neurobiological mechanisms underlying EMDR, and in part to our incomplete understanding of PTSD. In order to model PTSD, we used a biologically inspired computational model based on firing rate units, encompassing the cortex, hippocampus, and amygdala. Through the modulation of its parameters, we fitted real data from patients treated with EMDR or classical exposure therapy. This allowed us to gain insights into PTSD mechanisms and to investigate how EMDR achieves trauma remission.

A computational model of inner speech supporting flexible goal-directed behaviour in Autism.

Experimental and computational studies propose that inner speech boosts categorisation skills and executive functions, making human behaviour more focused and flexible. In addition, many clinical studies highlight a relationship between poor inner-speech and an executive impairment in autism spectrum condition (ASC), but contrasting findings are reported. Here we directly investigate the latter issue through a previously implemented and validated computational model of the Wisconsin Cards Sorting Tests. In particular, the model was applied to explore potential individual differences in cognitive flexibility and inner speech contribution in autistic and neurotypical participants. Our model predicts that the use of inner-speech could increase along the life-span of neurotypical participants but would be reduced in autistic ones. Although we found more attentional failures (i.e., wrong behavioural rule switches) in autistic children/teenagers and more perseverative behaviours in autistic young/older adults, only autistic children and older adults exhibited a lower performance (i.e., fewer consecutive correct rule switches) than matched control groups. Overall, our results corroborate the idea that the reduced use of inner speech could represent a disadvantage for autistic children and autistic older adults. Moreover, the results suggest that cognitive-behavioural therapies should focus on developing inner speech skills in autistic children as this could provide cognitive support throughout their whole life span.

Integrating unsupervised and reinforcement learning in human categorical perception: A computational model.

Categorical perception identifies a tuning of human perceptual systems that can occur during the execution of a categorisation task. Despite the fact that experimental studies and computational models suggest that this tuning is influenced by task-independent effects (e.g., based on Hebbian and unsupervised learning, UL) and task-dependent effects (e.g., based on reward signals and reinforcement learning, RL), no model studies the UL/RL interaction during the emergence of categorical perception. Here we have investigated the effects of this interaction, proposing a system-level neuro-inspired computational architecture in which a perceptual component integrates UL and RL processes. The model has been tested with a categorisation task and the results show that a balanced mix of unsupervised and reinforcement learning leads to the emergence of a suitable categorical perception and the best performance in the task. Indeed, an excessive unsupervised learning contribution tends to not identify task-relevant features while an excessive reinforcement learning contribution tends to initially learn slowly and then to reach sub-optimal performance. These results are consistent with the experimental evidence regarding categorical activations of extrastriate cortices in healthy conditions. Finally, the results produced by the two extreme cases of our model can explain the existence of several factors that may lead to sensory alterations in autistic people.

Neural Circuits Underlying Social Fear in Rodents: An Integrative Computational Model.

Social avoidance in rodents arises from a complex interplay between the prefrontal cortex and subcortical structures, such as the ventromedial hypothalamus and the dorsal periaqueductal gray matter. Experimental studies are revealing the contribution of these areas, but an integrative view and model of how they interact to produce adaptive behavior are still lacking. Here, we present a computational model of social avoidance, proposing a set of integrated hypotheses on the possible macro organization of the brain system underlying this phenomenon. The model is validated by accounting for several different empirical findings and produces predictions to be tested in future experiments.

A Computational Model Integrating Multiple Phenomena on Cued Fear Conditioning, Extinction, and Reinstatement.

Conditioning, extinction, and reinstatement are fundamental learning processes of animal adaptation, also strongly involved in human pathologies such as post-traumatic stress disorder, anxiety, depression, and dependencies. Cued fear conditioning, extinction, restatement, and systematic manipulations of the underlying brain amygdala and medial prefrontal cortex, represent key experimental paradigms to study such processes. Numerous empirical studies have revealed several aspects and the neural systems and plasticity underlying them, but at the moment we lack a comprehensive view. Here we propose a computational model based on firing rate leaky units that contributes to such integration by accounting for 25 different experiments on fear conditioning, extinction, and restatement, on the basis of a single neural architecture having a structure and plasticity grounded in known brain biology. This allows the model to furnish three novel contributions to understand these open issues: (a) the functioning of the central and lateral amygdala system supporting conditioning; (b) the role played by the endocannabinoids system in within- and between-session extinction; (c) the formation of three important types of neurons underlying fear processing, namely fear, extinction, and persistent neurons. The model integration of the results on fear conditioning goes substantially beyond what was done in previous models.

Enhancing Plasticity Mechanisms in the Mouse Motor Cortex by Anodal Transcranial Direct-Current Stimulation: The Contribution of Nitric Oxide Signaling.

Consistent body of evidence shows that transcranial direct-current stimulation (tDCS) over the primary motor cortex (M1) facilitates motor learning and promotes recovery after stroke. However, the knowledge of molecular mechanisms behind tDCS effects needs to be deepened for a more rational use of this technique in clinical settings. Here we characterized the effects of anodal tDCS of M1, focusing on its impact on glutamatergic synaptic transmission and plasticity. Mice subjected to tDCS displayed increased long-term potentiation (LTP) and enhanced basal synaptic transmission at layer II/III horizontal connections. They performed better than sham-stimulated mice in the single-pellet reaching task and exhibited increased forelimb strength. Dendritic spine density of layer II/III pyramidal neurons was also increased by tDCS. At molecular level, tDCS enhanced: 1) BDNF expression, 2) phosphorylation of CREB, CaMKII, and GluA1, and 3) S-nitrosylation of GluA1 and HDAC2. Blockade of nitric oxide synthesis by L-NAME prevented the tDCS-induced enhancement of GluA1 phosphorylation at Ser831 and BDNF levels, as well as of miniature excitatory postsynaptic current (mEPSC) frequency, LTP and reaching performance. Collectively, these findings demonstrate that anodal tDCS engages plasticity mechanisms in the M1 and highlight a role for nitric oxide (NO) as a novel mediator of tDCS effects.

Glutamate/GABA co-release selectively influences postsynaptic glutamate receptors in mouse cortical neurons.

Cultured rat cortical neurons co-expressing VGLUT1 and VGAT (mixed synapses) co-release Glu and GABA. Here, mixed synapses were studied in cultured mouse cortical neurons to verify whether in mice mixed synapses co-release Glu and GABA, and to gain insight into how they may influence excitation/inhibition balance. Results showed the existence of synapses and autapses that co-release Glu and GABA in cultured mouse cortical neurons, and the ability of both neurotransmitters to evoke postsynaptic responses mediated by ionotropic receptors. We studied the short-term plasticity of glutamatergic, GABAergic, and mixed responses and we found that the kinetics of mixPSC amplitude depression was similar to that observed in EPSCs, but it was different from that of IPSCs. We found similar presynaptic release characteristics in glutamatergic and mixed synapses. Analysis of postsynaptic features, obtained by measuring AMPAR- and NMDAR-mediated currents, showed that AMPAR-mediated currents were significantly higher in pure glutamatergic than in mixed synapses, whereas NMDAR-mediated currents were not significantly different from those measured in mixed synapses. Overall, our findings demonstrate that glutamatergic and mixed synapses share similar electrophysiological properties. However, co-release of GABA and Glu influences postsynaptic ionotropic glutamatergic receptor subtypes, thus selectively influencing AMPAR-mediated currents. These findings strengthen the view that mixed neurons can play a key role in CNS development and in maintaining the excitation-inhibition balance.

GSK3ß Modulates Timing-Dependent Long-Term Depression Through Direct Phosphorylation of Kv4.2 Channels.

Spike timing-dependent plasticity (STDP) is a form of activity-dependent remodeling of synaptic strength that underlies memory formation. Despite its key role in dictating learning rules in the brain circuits, the molecular mechanisms mediating STDP are still poorly understood. Here, we show that spike timing-dependent long-term depression (tLTD) and A-type K+ currents are modulated by pharmacological agents affecting the levels of active glycogen-synthase kinase 3 (GSK3) and by GSK3ß knockdown in layer 2/3 of the mouse somatosensory cortex. Moreover, the blockade of A-type K+ currents mimics the effects of GSK3 up-regulation on tLTD and occludes further changes in synaptic strength. Pharmacological, immunohistochemical and biochemical experiments revealed that GSK3ß influence over tLTD induction is mediated by direct phosphorylation at Ser-616 of the Kv4.2 subunit, a molecular determinant of A-type K+ currents. Collectively, these results identify the functional interaction between GSK3ß and Kv4.2 channel as a novel mechanism for tLTD modulation providing exciting insight into the understanding of GSK3ß role in synaptic plasticity.

Brain insulin resistance impairs hippocampal synaptic plasticity and memory by increasing GluA1 palmitoylation through FoxO3a.

High-fat diet (HFD) and metabolic diseases cause detrimental effects on hippocampal synaptic plasticity, learning, and memory through molecular mechanisms still poorly understood. Here, we demonstrate that HFD increases palmitic acid deposition in the hippocampus and induces hippocampal insulin resistance leading to FoxO3a-mediated overexpression of the palmitoyltransferase zDHHC3. The excess of palmitic acid along with higher zDHHC3 levels causes hyper-palmitoylation of AMPA glutamate receptor subunit GluA1, hindering its activity-dependent trafficking to the plasma membrane. Accordingly, AMPAR current amplitudes and, more importantly, their potentiation underlying synaptic plasticity were inhibited, as well as hippocampal-dependent memory. Hippocampus-specific silencing of Zdhhc3 and, interestingly enough, intranasal injection of the palmitoyltransferase inhibitor, 2-bromopalmitate, counteract GluA1 hyper-palmitoylation and restore synaptic plasticity and memory in HFD mice. Our data reveal a key role of FoxO3a/Zdhhc3/GluA1 axis in the HFD-dependent impairment of cognitive function and identify a novel mechanism underlying the cross talk between metabolic and cognitive disorders.

Loss of Leptin-Induced Modulation of Hippocampal Synaptic Trasmission and Signal Transduction in High-Fat Diet-Fed Mice.

Hippocampal plasticity is triggered by a variety of stimuli including sensory inputs, neurotrophins and inflammation. Leptin, whose primary function is to regulate food intake and energy expenditure, has been recently shown to affect hippocampal neurogenesis and plasticity. Interestingly, mice fed a high-fat diet (HFD) exhibit impaired hippocampal function, but the underlying mechanisms are poorly understood. To address this issue, we compared leptin responsiveness of hippocampal neurons in control and HFD-fed mice by combining single-cell electrophysiology and biochemical assays. We found that leptin modulated spontaneous and evoked synaptic transmission in control, but not HFD, mice. This functional impairment was paralleled by blunted activation of STAT-3, one of the key signal transduction pathways controlled by the fully functional isoform of the leptin receptor, ObRb. In addition, SOCS-3 expression was non-responsive to leptin, indicating that modulation of negative feedback impinging on ObRb was also altered. Our results advance the understanding of leptin action on hippocampal plasticity and, more importantly, suggest that leptin resistance is a key determinant of hippocampal dysfunction associated with hypercaloric diet.

Tis21 is required for adult neurogenesis in the subventricular zone and for olfactory behavior regulating cyclins, BMP4, Hes1/5 and Ids.

Bone morphogenic proteins (BMPs) and the Notch pathway regulate quiescence and self-renewal of stem cells of the subventricular zone (SVZ), an adult neurogenic niche. Here we analyze the role at the intersection of these pathways of Tis21 (Btg2/PC3), a gene regulating proliferation and differentiation of adult SVZ stem and progenitor cells. In Tis21-null SVZ and cultured neurospheres, we observed a strong decrease in the expression of BMP4 and its effectors Smad1/8, while the Notch anti-neural mediators Hes1/5 and the basic helix-loop-helix (bHLH) inhibitors Id1-3 increased. Consistently, expression of the proneural bHLH gene NeuroD1 decreased. Moreover, cyclins D1/2, A2, and E were strongly up-regulated. Thus, in the SVZ Tis21 activates the BMP pathway and inhibits the Notch pathway and the cell cycle. Correspondingly, the Tis21-null SVZ stem cells greatly increased; nonetheless, the proliferating neuroblasts diminished, whereas the post-mitotic neuroblasts paradoxically accumulated in SVZ, failing to migrate along the rostral migratory stream to the olfactory bulb. The ability, however, of neuroblasts to migrate from SVZ explants was not affected, suggesting that Tis21-null neuroblasts do not migrate to the olfactory bulb because of a defect in terminal differentiation. Notably, BMP4 addition or Id3 silencing rescued the defective differentiation observed in Tis21-null neurospheres, indicating that they mediate the Tis21 pro-differentiative action. The reduced number of granule neurons in the Tis21-null olfactory bulb led to a defect in olfactory detection threshold, without effect on olfactory memory, also suggesting that within olfactory circuits new granule neurons play a primary role in odor sensitivity rather than in memory.

Running rescues defective adult neurogenesis by shortening the length of the cell cycle of neural stem and progenitor cells.

Physical exercise increases the generation of new neurons in adult neurogenesis. However, only few studies have investigated the beneficial effects of physical exercise in paradigms of impaired neurogenesis. Here, we demonstrate that running fully reverses the deficient adult neurogenesis within the hippocampus and subventricular zone of the lateral ventricle, observed in mice lacking the antiproliferative gene Btg1. We also evaluated for the first time how running influences the cell cycle kinetics of stem and precursor subpopulations of wild-type and Btg1-null mice, using a new method to determine the cell cycle length. Our data show that in wild-type mice running leads to a cell cycle shortening only of NeuroD1-positive progenitor cells. In contrast, in Btg1-null mice, physical exercise fully reactivates the defective hippocampal neurogenesis, by shortening the S-phase length and the overall cell cycle duration of both neural stem (glial fibrillary acidic protein(+) and Sox2(+)) and progenitor (NeuroD1(+)) cells. These events are sufficient and necessary to reactivate the hyperproliferation observed in Btg1-null early-postnatal mice and to expand the pool of adult neural stem and progenitor cells. Such a sustained increase of cell proliferation in Btg1-null mice after running provides a long-lasting increment of proliferation, differentiation, and production of newborn neurons, which rescues the impaired pattern separation previously identified in Btg1-null mice. This study shows that running positively affects the cell cycle kinetics of specific subpopulations of newly generated neurons and suggests that the plasticity of neural stem cells without cell cycle inhibitory control is reactivated by running, with implications for the long-term modulation of neurogenesis.