Biology subjects

Paik, S.-B.

Publications and source records attributed to Paik, S.-B..

At least 19 recordsLinked to original sources

bioRxiv2025

Gradual sensory maturation promotes abstract representation learning

Human infants begin life with limited visual capacities, such as low acuity and poor color sensitivity, due to gradual sensory maturation. In contrast, machine learning models are trained on high-fidelity inputs from the outset, often leading to shortcut learning and overfitting to spurious correlations. Here, we show that early sensory immaturity plays a critical role in shaping bias-resistant, abstract visual representations that conventional models struggle to develop. Using neural network simulations and human psychophysics experiments, we demonstrate that gradual sensory development supports the emergence of robust and generalizable internal representations, reduces reliance on superficial cues, and promotes disentangled representations that enable compositional reconstruction and visual imagination. Comparative analyses of human and model behavior reveal shared patterns of bias resistance and adaptive generalization, including resilience to misleading information. Our findings suggest that gradual sensory maturation is not merely a developmental constraint, but rather a key mechanism that enables abstract representation learning. One sentence summaryEarly-stage sensory immaturity guides the development of abstract, bias-resistant representations that enable generalization and compositionality, providing a functional account of gradual sensory maturation. HighlightsO_LIThe essential yet overlooked role of gradual sensory maturation was explored C_LIO_LIEarly sensory immaturity promotes abstract representations resistant to shortcut learning C_LIO_LIEmergent representations support compositional reconstruction of novel visual attributes C_LIO_LIHuman and model behaviors show similar bias resistance and adaptive generalization C_LI

neuroscience↗

bioRxiv2024

Hard-wired visual filters for environment-agnostic object recognition

Conventional deep neural networks (DNNs) are highly susceptible to variations in input domains, unlike biological brains which effectively adapt to environmental changes. Here, we demonstrate that hard-wired Gabor filters, replicating the structure of receptive fields in the brains early visual pathway, facilitate environment-agnostic object recognition without overfitting. Our approach involved fixing the pre-designed Gabor filters in the early layers of DNNs, preventing any alterations during training. Despite the restricted learning flexibility of this model, our networks maintained robust performance even under significant domain shifts, in contrast to conventional DNNs that typically fail in similar conditions. We found that our model effectively clustered identical "classes" across diverse domains, while conventional DNNs tend to cluster images by "domain" in the latent space. We observed that the fixed Gabor filters enabled networks to encode global shape information rather than local texture features, thereby mitigating the risk of overfitting. One sentence summaryHard-wired Gabor filters enable environment-agnostic object recognition without overfitting. Research HighlightsO_LIConventional deep neural networks (DNNs) are vulnerable to input domain variations C_LIO_LIHard-wired Gabor filters facilitate environment-agnostic object recognition C_LIO_LIFixed Gabor filters prevent overfitting and facilitate shape-based classifications C_LIO_LIOur model cluster identical "classes" while conventional DNNs cluster by "domain" C_LI

neuroscience↗

bioRxiv2023

Hierarchical gradient of timescales in the mammalian forebrain

Many anatomical and physiological features of cortical circuits, ranging from the biophysical properties of synapses to the connectivity patterns among different neuron types, exhibit consistent variation along the hierarchical axis from sensory to association areas. Notably, the scale of temporal correlation of neural activity at rest, known as the intrinsic timescale, increases systematically along this hierarchy in both primates and rodents, analogous to the growing scale and complexity of spatial receptive fields. However, how the timescales for task-related activity vary across brain regions and whether their hierarchical organization appears consistently across different mammalian species remain unexplored. Here, we show that both the intrinsic timescale and the timescales of task-related activity follow a similar hierarchical gradient in the cortices of monkeys, rats, and mice. We also found that these timescales covary similarly in both the cortex and basal ganglia, whereas the timescales of thalamic activity are shorter than cortical timescales and do not conform to the hierarchical order predicted by their cortical projections. These results suggest that the hierarchical gradient of cortical timescales might be a universal feature of intra-cortical circuits in the mammalian brain. Significance StatementA gradual increase in the intrinsic timescales of cortical activity along the anatomical hierarchy reflects the functional specialization of cortical circuits. However, it is unknown whether this gradient of timescales is a common feature across distinct mammalian species in both intrinsic and task-related timescales and whether it is also observed in subcortical areas. This study reveals that the hierarchical gradient of multiple cortical timescales is conserved across multiple mammalian species. By contrast, thalamic timescales were shorter than cortical timescales and did not follow the hierarchical order inferred from their cortical projections. These findings imply a crucial role of intra-cortical connections in structuring distinct temporal dynamics observed across the cortex.

neuroscience↗

bioRxiv2022

Invariance of object detection in untrained deep neural networks

The ability to perceive visual objects with various types of transformations, such as rotation, translation, and scaling, is crucial for consistent object recognition. In machine learning, invariant object detection for a network is often implemented by augmentation with a massive number of training images, but the mechanism of invariant object detection in biological brains -- how invariance arises initially and whether it requires visual experience -- remains elusive. Here, using a model neural network of the hierarchical visual pathway of the brain, we show that invariance of object detection can emerge spontaneously in the complete absence of learning. First, we found that units selective to a particular object class arise in randomly initialized networks even before visual training. Intriguingly, these units show robust tuning to images of each object class under a wide range of image transformation types, such as viewpoint rotation. We confirmed that this "innate" invariance of object selectivity enables untrained networks to perform an object-detection task robustly, even with images that have been significantly modulated. Our computational model predicts that invariant object tuning originates from combinations of non-invariant units via random feedforward projections, and we confirmed that the predicted profile of feedforward projections is observed in untrained networks. Our results suggest that invariance of object detection is an innate characteristic that can emerge spontaneously in random feedforward networks. HighlightsO_LIObject-selective units spontaneously emerge in untrained deep neural networks C_LIO_LIObject selectivity maintains robustly in a wide range of image transformations C_LIO_LIFeedforward model can explain spontaneous emergence of the invariance C_LIO_LIInnate invariance enables invariant object detection without learning to variations C_LI

neuroscience↗

bioRxiv2022

Comparison of visual quantities in untrained deep neural networks

The ability to compare quantities of visual objects with two distinct measures, proportion and difference, is observed in newborn animals. Nevertheless, how this function originates in the brain, even before training, remains unknown. Here, we show that neuronal tuning for quantity comparison can arise spontaneously in completely untrained deep neural networks. Using a biologically inspired model neural network, we found that units selective to proportions and differences between visual quantities emerge in randomly initialized networks and that they enable the network to perform quantity comparison tasks. Further analysis shows that two distinct tunings to proportion and difference both originate from a random summation of monotonic, nonlinear responses to changes in relative quantities. Notably, we found that a slight difference in the nonlinearity profile determines the type of measure. Our results suggest that visual quantity comparisons are primitive types of functions that can emerge spontaneously in random feedforward networks. One sentence summaryThe ability to compare visual quantities arises spontaneously in untrained deep neural networks. Research HighlightsO_LIThe ability to compare visual quantity arises spontaneously in untrained networks C_LIO_LIDistinct tunings to measure proportion and difference of quantities are observed C_LIO_LIRandom wiring of monotonic, nonlinear activity induces quantity-comparison units C_LIO_LIThe nonlinearity pattern of the source unit determines the type of target measure C_LI

neuroscience↗

bioRxiv2022

Npas4-mediated dopaminergic regulation of fear memory states

Amygdala circuitry encodes associations between conditioned stimuli and aversive unconditioned stimuli, and also controls fear expression (Pape and Pare, 2010). However, whether and how irrelevant information for unpaired conditioned stimuli (CS-) is discretely processed, and how it was influenced by stress remain unknown. CS- memory is retrievable immediately after fear conditioning, but then becomes silent after memory consolidation in mice. Synaptic pathway from the lateral to the anterior basal amygdala gates the expression of CS- memory, depending upon Npas4-mediated Drd4 synthesis. The upregulation of Npas4-Drd4 axis, which is precluded by corticosterone, shifts functional states of neural engrams for CS- memory toward silent states and limits its retrievability. In here, we demonstrate the cellular and molecular mechanisms regulating the functional states of neural engrams, which can be switched or maintained, supporting discriminative memory.

neuroscience↗

bioRxiv2021

Suppressed prefrontal neuronal firing variability and impaired social representation in IRSp53-mutant mice

Social deficit is a major feature of neuropsychiatric disorders, including autism spectrum disorders, schizophrenia, and attention-deficit/hyperactivity disorder, but its neural mechanisms remain unclear. Here, we examined neuronal discharge characteristics in the medial prefrontal cortex (mPFC) of IRSp53-mutant mice, which show social deficits, during social approach. IRSp53-mutant excitatory mPFC neurons displayed an increase in baseline neuronal firing and decreases in variability and dynamic range of firing rates and burst firing during social and non-social target approaches compared to wild-type controls. As a consequence, their firing activity was less differential between social and non-social targets. In addition, there was a decrease in the proportion of excitatory mPFC neurons encoding social information but not that of those encoding non-social information. These results suggest that insufficient neuronal activity dynamics may underlie impaired cortical encoding of social information and social behaviors in IRSp53-mutant mice.

neuroscience↗

bioRxiv2021

Parallel processing of working memory and temporal information by distinct types of cortical projection neurons

It is unclear how different types of cortical projection neurons work together to support diverse cortical functions. We examined the discharge characteristics and inactivation effects of intratelencephalic (IT) and pyramidal tract (PT) neurons--two major types of cortical excitatory neurons that project to cortical and subcortical structures, respectively--in the medial prefrontal cortex of mice performing a delayed response task. We found that IT neurons, but not PT neurons, convey significant working memory-related signals. We also found that the inactivation of IT neurons, but not PT neurons, impairs behavioral performance. In contrast, PT neurons convey far more temporal information than IT neurons during the delay period. Our results indicate a division of labor between IT and PT projection neurons in the prefrontal cortex for the maintenance of working memory and for tracking the passage of time, respectively.

neuroscience↗

bioRxiv2020

Slow rhythmic eye motion predicts periodic alternation of bistable perception

Bistable perception is characterized by periodic alternation between two different perceptual interpretations, the mechanism of which is poorly understood. Herein, we show that perceptual decisions in bistable perception are strongly correlated with slow rhythmic eye motion, the frequency of which varies across individuals. From eye gaze trajectory measurements during three types of bistable tasks, we found that each subjects gaze position oscillates slowly(less than 1Hz), and that this frequency matches that of bistable perceptual alternation. Notably, the motion of the eye apparently moves in opposite directions before two opposite perceptual decisions, and this enables the prediction of the timing and direction of perceptual alternation from eye motion. We also found that the correlation between eye movement and a perceptual decision is maintained during variations of the alternation frequency by the intentional switching or retaining of perceived states. This result suggests that periodic bistable perception is phase-locked with rhythmic eye motion.

neuroscience↗

bioRxiv2020

Sparse long-range connections in visual cortex for cost-efficient small-world networks

The brain performs visual object recognition using much shallower hierarchical stages than artificial deep neural networks employ. However, the mechanism underlying this cost-efficient function is elusive. Here, we show that cortical long-range connectivity(LRC) may enable this parsimonious organization of circuits for balancing cost and performance. Using model network simulations based on data in tree shrews, we found that sparse LRCs, when added to local connections, organize a small-world network that dramatically enhances object recognition of shallow feedforward networks. We found that optimization of the ratio between LRCs and local connections maximizes the small-worldness and task performance of the network, by minimizing the total length of wiring needed for integration of the global information. We also found that the effect of LRCs varies by network size, which explains the existence of species-specific LRCs in mammalian visual cortex of various sizes. Our results demonstrate a biological strategy to achieve cost-efficient brain circuits. HighlightsO_LILong-range connections (LRCs) enhance the object recognition of shallow networks C_LIO_LISparse LRCs added to dense local connections organize a small-world type network C_LIO_LISmall-worldness of networks modulates the balance between performance and wiring cost C_LIO_LIDistinct LRCs in various species are due to the size-dependent effect of LRCs C_LI Significance statementThe hierarchical depth of the visual pathway in the brain is constrained by biological factors, whereas artificial deep neural networks consist of super-deep structures (i.e., as deep as computational power allows). Here, we show that long-range horizontal connections (LRCs) observed in mammalian visual cortex may enable shallow biological networks to perform cognitive tasks that require deeper artificial structures, by implementing cost-efficient organization of circuitry. Using model simulations based on anatomical data, we found that sparse LRCs, when added to dense local circuits, organize "small-world" type networks and that this dramatically enhances image classification performance by integrating both local and global components of visual stimulus. Our findings show a biological strategy of brain circuitry to balance sensory performance and wiring cost in the networks. One sentence summaryCortical long-range connections organize a small-world type network to achieve cost-efficient functional circuits under biological constraints

neuroscience↗

bioRxiv2020

Spontaneous retinal waves generate long-range horizontal connectivity in visual cortex

In the primary visual cortex (V1) of higher mammals, long-range horizontal connections (LHCs) are observed to develop, linking iso-orientation domains of cortical tuning. It is unknown how this feature-specific wiring of circuitry develops before eye opening. Here, we show that LHCs in V1 may originate from spatio-temporally structured feedforward activities generated from spontaneous retinal waves. Using model simulations based on the anatomy and observed activity patterns of the retina, we show that waves propagating in retinal mosaics can initialize the wiring of LHCs by co-activating neurons of similar tuning, whereas equivalent random activities cannot induce such organizations. Simulations showed that emerged LHCs can produce the patterned activities observed in V1, matching topography of the underlying orientation map. We also confirmed that the model can also reproduce orientation-specific microcircuits in salt-and-pepper organizations in rodents. Our results imply that early peripheral activities contribute significantly to cortical development of functional circuits. HighlightsO_LIDevelopmental model of long-range horizontal connections (LHCs) in V1 is simulated C_LIO_LISpontaneous retinal waves generate feature-specific wiring of LHCs in visual cortex C_LIO_LIEmerged LHCs induce orientation-matching patterns of spontaneous cortical activity C_LIO_LIRetinal waves induce orientation-specific microcircuits of visual cortex in rodents C_LI Significance statementLong-range horizontal connections (LHCs) in the primary visual cortex (V1) are observed to emerge before the onset of visual experience, selectively connecting iso-domains of orientation maps. However, it is unknown how such tuning-specific wirings develop before eye-opening. Here, we show that LHCs in V1 originate from the tuning-specific activation of cortical neurons by spontaneous retinal waves during early developmental stages. Our simulations of a visual cortex model show that feedforward activities from the retina initialize the spatial organization of activity patterns in V1, which induces visual feature-specific wirings of V1 neurons. Our model also explains the origin of cortical microcircuits observed in rodents, suggesting that the proposed developmental mechanism is applicable universally to circuits of various mammalian species.

neuroscience↗

bioRxiv2020

Projection of orthogonal tiling from the retina to the visual cortex

In higher mammals, the primary visual cortex (V1) is organized into diverse tuning maps of visual features such as orientation, spatial frequency and ocular dominance. The topography of these maps is observed to intersect orthogonally, implying that a developmental principle for efficient tiling of sensory modules may exist. However, it remains unclear how such a systematic relationship among cortical tuning maps could develop. Here, we show that the orthogonal organization of tuning modules already exist in retinal ganglion cell (RGC) mosaics, and that this provides a blueprint of the orthogonal organization in V1. Firstly, from the analysis of multi-electrode recording data in V1, we found that the ON-OFF subregion distance of receptive fields varies periodically across the cortical surface, strongly correlated to ocular dominance and spatial frequency in the area. Further, the ON-OFF alignment angle, that is orthogonal to the ON-OFF distance, appears to correlate with orientation tuning. These suggest that the orthogonal organization in V1 may originate from the spatial organization of the ON-OFF receptive fields in the bottom-up projections, and this scenario was tested from analysis of the RGC mosaics data in monkeys and cats. We found that the ON-OFF RGC distance and ON-OFF angle of neighbouring RGCs are organized into a topographic tiling across mosaics, analogous to the orthogonal intersection of cortical tuning maps. These findings suggest that the regularly structured ON-OFF patterns mirrored from a retina may initiate efficient tiling of functional domains in V1. HighlightsO_LIOrthogonal organization of visual tuning maps are observed in both V1 and the retina C_LIO_LICortical tuning maps are correlated with the profile of ON-OFF feedforward projections C_LIO_LIThe profile of ON-OFF receptive fields varies periodically across the V1 surface C_LIO_LIRegularly structured RGC patterns initiate the orthogonal tiling of sensory modules in V1 C_LI

neuroscience↗

bioRxiv2019

Spontaneous generation of face recognition in untrained deep neural networks

Face-selective neurons are observed in the primate visual pathway and are considered the basis of facial recognition in the brain. However, it is debated whether this neuronal selectivity can arise spontaneously, or requires training from visual experience. Here, we show that face-selective neurons arise spontaneously in random feedforward networks in the absence of learning. Using biologically inspired deep neural networks, we found that face-selective neurons arise under three different network conditions: one trained using non-face natural images, one randomized after being trained, and one never trained. We confirmed that spontaneously emerged face-selective neurons show the biological view-point-invariant characteristics observed in monkeys. Such neurons suddenly vanished when feedforward weight variation declined to a certain level. Our results suggest that innate face-selectivity originates from statistical variation of the feedforward projections in hierarchical neural networks.

neuroscience↗

bioRxiv2019

Spontaneous generation of innate number sense in untrained deep neural networks

Number-selective neurons are observed in numerically naive animals, but it was not understood how this innate function emerges in the brain. Here, we show that neurons tuned to numbers can arise in random feedforward networks, even in the complete absence of learning. Using a biologically inspired deep neural network, we found that number tuning arises in three cases of networks: one trained to non-numerical natural images, one randomized after trained, and one never trained. Number-tuned neurons showed characteristics that were observed in the brain following the Weber-Fechner law. These neurons suddenly vanished when the feedforward weight variation decreased to a certain level. These results suggest that number tuning can develop from the statistical variation of bottom-up projections in the visual pathway, initializing innate number sense.

neuroscience↗

bioRxiv2019

Simultaneous development and periodic clustering of simple and complex cells in visual cortex

Neurons in the primary visual cortex (V1) are often classified as simple or complex cells, but it is debated whether they are discrete hierarchical classes of neurons developing sequentially, or if they represent a continuum of variation within a single class of cells developing simultaneously. Herein, we show that simple and complex cells may arise simultaneously from the universal process of retinal development. From analysis of the cortical receptive fields in cats, we show evidence that simple and complex cells originate from the periodic variation of ON-OFF segregation in the feedforward projection of retinal mosaics, by which they organize into periodic clusters in V1. Our key prediction that clusters of simple and complex cells correlate topographically with orientation maps was confirmed by data in cats. Our results suggest that simple and complex cells are not two distinct neural populations but arise from common retinal afferents, simultaneous with orientation tuning.\n\nHighlightsO_LISimple and complex cells arise simultaneously from retinal afferents.\nC_LIO_LISimple/complex cells are organized into periodic clusters across visual cortex.\nC_LIO_LISimple/complex clusters are topographically correlated with orientation maps.\nC_LIO_LIDevelopment of clustered cells in V1 is explained by the Paik-Ringach model.\nC_LI

neuroscience↗

bioRxiv2019

Classification of columnar and salt-and-pepper organization in mammalian visual cortex

In mammalian visual cortex, neural tuning to stimulus orientation is organized in either columnar1 or salt- and-pepper2 patterns across species. This is often considered to reflect disparate mechanisms of cortical development across mammalian taxa. However, it is unknown whether different cortical architectures are generated by species-specific mechanisms3,4, or simply originate from the variation of biological parameters within a universal principle of development5-8. We analysed neural parameters in eight mammalian species and found that cortical organization is predictable by a single factor: the retino-cortical mapping ratio. We show that a Nyquist sampling model explains parametric division of the patterns with high accuracy and that simulations of controlled mapping conditions reproduce both types of organization. Our results explain the origin of distinct cortical circuits under a universal development process.

neuroscience↗

bioRxiv2019

Distinct role of flexible and stable encoding in sequential working memory

The serial-position effect in working memory is considered important for studying how a sequence of sensory information can be retained and manipulated simultaneously in neural memory circuits. Here, via a precise analysis of the primacy and recency effects in human psychophysical experiments, we propose that stable and flexible coding take distinct roles of retaining and updating information in working memory, and that their combination induces serial-position effects spontaneously. We found that stable encoding retains memory to induce the primacy effect, while flexible encoding used for learning new inputs induces the recency effect. A model simulation based on human data, confirmed that a neural network with both flexible and stable synapses could reproduce the major characteristics of serial-position effects. Our new prediction, that the control of resource allocation by flexible-stable coding balance can modulate memory performance in sequence-specific manner, was supported by pre-cued memory performance data in humans.

neuroscience↗

bioRxiv2018

Automated 3-D mapping of single neurons in the standard brain atlas using single brain slices

Recent breakthroughs in neuroanatomical tracing methods have helped unravel complicated neural connectivity in whole brain tissue at a single cellular resolution. However, analysis of brain images remains dependent on highly subjective manual processing. In the present study, we introduce AMaSiNe, a novel software for automated mapping of single neurons in the standard mouse brain atlas. The AMaSiNe automatically calibrates alignment angles of each brain slice to match the Allen Reference Atlas (ARA), locates labeled neurons from multiple brain samples in a common brain space, and achieves a standardized 3D-rendered brain. Due to the high fidelity and reliability of AMaSiNe, the retinotopic structures of neural projections to the primary visual cortex (VISp) were determined from single and dual injections of the rabies virus onto different visual areas. Our results demonstrate that distinct retinotopic organization of bottom-up and top-down projections could be precisely mapped using AMaSiNe.

neuroscience↗