Attention activates winner-take-all competition among visual filters

Analyzing top-down visual attention in the context of gamma oscillations: a layer- dependent network-of- networks approach

Top-down visual attention is a fundamental cognitive process that allows individuals to selectively attend to salient visual stimuli in the environment. Recent empirical findings have revealed that gamma oscillations participate in the modulation of visual attention. However, computational studies face challenges when analyzing the attentional process in the context of gamma oscillation due to the unstable nature of gamma oscillations and the complexity induced by the layered fashion in the visual cortex. In this study, we propose a layer-dependent network-of-networks approach to analyze such attention with gamma oscillations. The model is validated by reproducing empirical findings on orientation preference and the enhancement of neuronal response due to top-down attention. We perform parameter plane analysis to classify neuronal responses into several patterns and find that the neuronal response to sensory and attention signals was modulated by the heterogeneity of the neuronal population. Furthermore, we revealed a counter-intuitive scenario that the excitatory populations in layer 2/3 and layer 5 exhibit opposite responses to the attentional input. By modification of the original model, we confirmed layer 6 plays an indispensable role in such cases. Our findings uncover the layer-dependent dynamics in the cortical processing of visual attention and open up new possibilities for further research on layer-dependent properties in the cerebral cortex.

Similarity in motion binds and bends judgments of aspect ratio

It is well known that objects become grouped in perceptual organization when they share some visual feature, like a common direction of motion. Less well known is that grouping can change how people perceive a set of objects. For example, when a pair of shapes consistently share a common region of space, their aspect ratios tend to be perceived as more similar (are attracted toward each other). Conversely, when shapes are assigned to different regions in space their aspect ratios repel from each other. Here we examine whether the visual system produce both attractive and repulsive distortions when the state of grouping between a pair of shapes changes on a moment-to-moment basis. Observers viewed a pair of ellipses that differed in terms of how flat or tall they were and reported the aspect ratio of one ellipse from the pair. Each ellipse was defined by a cloud of coherently-moving dots, and the dots within the two ellipses had either the same or different directions of motion, varying from trial-to-trial. We found that the cued ellipse's aspect ratio was reported to be repelled from the aspect ratio of the uncued ellipse when the shapes had different directions of motion compared to when they had the same direction of motion. These results suggest that the visual system can adaptively alter visual experience based on grouping, in particular, repelling the appearance of objects when they do not appear to go together, and it can do so quickly and flexibly.

Implicit Selective Attention: The Role of the Mesencephalic-basal Ganglia System

The ability of the brain to recognize and orient attention to relevant stimuli appearing in the visual field is highlighted by a tuning process, which involves modulating the early visual system by both cortical and subcortical brain areas. Selective attention is coordinated not only by the output of stimulus-based saliency maps but is also influenced by top-down cognitive factors, such as internal states, goals, or previous experiences. The basal ganglia system plays a key role in implicitly modulating the underlying mechanisms of selective attention, favouring the formation and maintenance of implicit sensory-motor memories that are capable of automatically modifying the output of priority maps in sensory-motor structures of the midbrain, such as the superior colliculus. The article presents an overview of the recent literature outlining the crucial contribution of several subcortical structures to the processing of different sources of salient stimuli. In detail, we will focus on how the mesencephalic- basal ganglia closed loops contribute to implicitly addressing and modulating selective attention to prioritized stimuli. We conclude by discussing implicit behavioural responses observed in clinical populations in which awareness is compromised at some level. Implicit (emergent) awareness in clinical conditions that can be accompanied by manifest anosognosic symptomatology (i.e., hemiplegia) or involving abnormal conscious processing of visual information (i.e., unilateral spatial neglect and blindsight) represents interesting neurocognitive "test cases" for inferences about mesencephalicbasal ganglia closed-loops involvement in the formation of implicit sensory-motor memories.

Compact deep neural network models of visual cortex

A powerful approach to understanding the computations carried out in visual cortex is to develop models that predict neural responses to arbitrary images. Deep neural network (DNN) models have worked remarkably well at predicting neural responses [1, 2, 3], yet their underlying computations remain buried in millions of parameters. Have we simply replaced one complicated system in vivo with another in silico? Here, we train a data-driven deep ensemble model that predicts macaque V4 responses ~50% more accurately than currently-used task-driven DNN models. We then compress this deep ensemble to identify compact models that have 5,000x fewer parameters yet equivalent accuracy as the deep ensemble. We verified that the stimulus preferences of the compact models matched those of the real V4 neurons by measuring V4 responses to both 'maximizing' and adversarial images generated using compact models. We then analyzed the inner workings of the compact models and discovered a common circuit motif: Compact models share a similar set of filters in early stages of processing but then specialize by heavily consolidating this shared representation with a precise readout. This suggests that a V4 neuron's stimulus preference is determined entirely by its consolidation step. To demonstrate this, we investigated the compression step of a dot-detecting compact model and found a set of simple computations that may be carried out by dot-selective V4 neurons. Overall, our work demonstrates that the DNN models currently used in computational neuroscience are needlessly large; our approach provides a new way forward for obtaining explainable, high-accuracy models of visual cortical neurons.

The Architecture of Object-Based Attention

The allocation of attention to objects raises several intriguing questions: What are objects, how does attention access them, what anatomical regions are involved? Here, we review recent progress in the field to determine the mechanisms underlying object-based attention. First, findings from unconscious priming and cueing suggest that the preattentive targets of object-based attention can be fully developed object representations that have reached the level of identity. Next, the control of object-based attention appears to come from ventral visual areas specialized in object analysis that project downward to early visual areas. How feedback from object areas can accurately target the object's specific locations and features is unknown but recent work in autoencoding has made this plausible. Finally, we suggest that the three classic modes of attention may not be as independent as is commonly considered, and instead could all rely on object-based attention. Specifically, studies show that attention can be allocated to the separated members of a group-without affecting the space between them-matching the defining property of feature-based attention. At the same time, object-based attention directed to a single small item has the properties of space-based attention. We outline the architecture of object-based attention, the novel predictions it brings, and discuss how it works in parallel with other attention pathways.

Early sensory gain control is dominated by obligatory and global feature-based attention in top-down shifts of combined spatial and feature-based attention

What are the dynamics of global feature-based and spatial attention, when deployed together? In an attentional shifting experiment, flanked by three control experiments, we investigated neural temporal dynamics of combined attentional shifts. For this purpose, orange- and blue-frequency-tagged spatially overlapping Random Dot Kinematograms were presented in the left and right visual hemifield to elicit continuous steady-state-visual-evoked-potentials. After being initially engaged in a fixation cross task, participants were at some point in time cued to shift attention to one of the Random Dot Kinematograms, to detect and respond to brief coherent motion events, while ignoring all such events in other Random Dot Kinematograms. The analysis of steady-state visual-evoked potentials allowed us to map time courses and dynamics of early sensory-gain modulations by attention. This revealed a time-invariant amplification of the to-be attended color both at the attended and the unattended side, followed by suppression for the to-be-ignored color at attended and unattended sides. Across all experiments, global and obligatory feature-based selection dominated early sensory gain modulations, whereas spatial attention played a minor modulatory role. However, analyses of behavior and neural markers such as alpha-band activity and event-related potentials to target- and distractor-event processing, revealed clear modulations by spatial attention.

Visual attention to features and space in mice using reverse correlation

Visual attention allows the brain to evoke behaviors based on the most important visual features. Mouse models offer immense potential to gain a circuit-level understanding of this phenomenon, yet how mice distribute attention across features and locations is not well understood. Here, we describe a new approach to address this limitation by training mice to detect weak vertical bars in a background of dynamic noise while spatial cues manipulate their attention. By adapting a reverse-correlation method from human studies, we linked behavioral decisions to stimulus features and locations. We show that mice deployed attention to a small rostral region of the visual field. Within this region, mice attended to multiple features (orientation, spatial frequency, contrast) that indicated the presence of weak vertical bars. This attentional tuning grew with training, multiplicatively scaled behavioral sensitivity, approached that of an ideal observer, and resembled the effects of attention in humans. Taken together, we demonstrate that mice can simultaneously attend to multiple features and locations of a visual stimulus.

A dynamic 1/f noise protocol to assess visual attention without biasing perceptual processing

Psychophysical paradigms measure visual attention via localized test items to which observers must react or whose features have to be discriminated. These items, however, potentially interfere with the intended measurement, as they bias observers' spatial and temporal attention to their location and presentation time. Furthermore, visual sensitivity for conventional test items naturally decreases with retinal eccentricity, which prevents direct comparison of central and peripheral attention assessments. We developed a stimulus that overcomes these limitations. A brief oriented discrimination signal is seamlessly embedded into a continuously changing 1/f noise field, such that observers cannot anticipate potential test locations or times. Using our new protocol, we demonstrate that local orientation discrimination accuracy for 1/f filtered signals is largely independent of retinal eccentricity. Moreover, we show that items present in the visual field indeed shape the distribution of visual attention, suggesting that classical studies investigating the spatiotemporal dynamics of visual attention via localized test items may have obtained a biased measure. We recommend our protocol as an efficient method to evaluate the behavioral and neurophysiological correlates of attentional orienting across space and time.

Tilt aftereffect spreads across the visual field

The tilt aftereffect (TAE) is observed when adaptation to a tilted contour alters the perceived tilt of a subsequently presented contour. Thus far, TAE has been treated as a local aftereffect observed only at the location of the adapter. Whether and how TAE spreads to other locations in the visual field has not been systematically studied. Here, we sought an answer to this question by measuring TAE magnitudes at locations including but not limited to the adapter location. The adapter was a tilted grating presented at the same peripheral location throughout an experimental session. In a single trial, participants indicated the perceived tilt of a test grating presented after the adapter at one of fifteen locations in the same visual hemifield as the adapter. We found non-zero TAE magnitudes in all locations tested, showing that the effect spreads across the tested visual hemifield. Next, to establish a link between neuronal activity and behavioral results and to predict the possible neuronal origins of the spread, we built a computational model based on known characteristics of the visual cortex. The simulation results showed that the model could successfully capture the pattern of the behavioral results. Furthermore, the pattern of the optimized receptive field sizes suggests that mid-level visual areas, such as V4, could be critically involved in TAE and its spread across the visual field.

Hierarchy of Intra- and Cross-modal Redundancy Gains in Visuo-tactile Search: Evidence from the Posterior Contralateral Negativity

Redundant combination of target features from separable dimensions can expedite visual search. The dimension-weighting account explains these "redundancy gains" by assuming that the attention-guiding priority map integrates the feature-contrast signals generated by targets within the respective dimensions. The present study investigated whether this hierarchical architecture is sufficient to explain the gains accruing from redundant targets defined by features in different modalities, or whether an additional level of modality-specific priority coding is necessary, as postulated by the modality-weighting account (MWA). To address this, we had observers perform a visuo-tactile search task in which targets popped out by a visual feature (color or shape) or a tactile feature (vibro-tactile frequency) as well as any combination of these features. The RT gains turned out larger for visuo-tactile versus visual redundant targets, as predicted by the MWA. In addition, we analyzed two lateralized event-related EEG components: the posterior (PCN) and central (CCN) contralateral negativities, which are associated with visual and tactile attentional selection, respectively. The CCN proved to be a stable somatosensory component, unaffected by cross-modal redundancies. In contrast, the PCN was sensitive to cross-modal redundancies, evidenced by earlier onsets and higher amplitudes, which could not be explained by linear superposition of the earlier CCN onto the later PCN. Moreover, linear mixed-effect modeling of the PCN amplitude and timing parameters accounted for approximately 25% of the behavioral RT variance. Together, these behavioral and PCN effects support the hierarchy of priority-signal computation assumed by the MWA.

Global attentional selection of visual features is not associated with selective modulation of posterior alpha-band activity

Attending to a single feature, such as color or motion, leads to global modulation of neural processing associated with the representation of the attended features. Alpha-band modulations are hypothesized to be a marker (and even a mechanism) of the modulation of neural processing. By adopting a previously used attentional shifting paradigm, we examined whether alpha-band dynamics are linked to sustained Feature-Based-Attentional (FBA) selection. For this purpose, we presented task-irrelevant flickering random dot kinematograms (RDKs) in the periphery that either did or did not share the to-be-attended color of centrally presented task-relevant RDKs and should thus be subject to global FBA selection. Steady-state visual evoked potentials (SSVEPs) and alpha-band activity associated with these task-irrelevant RDKs were analyzed to quantify FBA modulation. Overall, the SSVEP results replicated previous findings: relative to a pre-cue baseline, SSVEP amplitudes for peripheral RDKs were significantly enhanced when these RDKs shared the to-be-attended color of the central RDKs and were not modulated when they shared the centrally to-be-ignored color. Nevertheless, there were no differences in alpha-band amplitude modulations between signals recorded contralateral to the RDKs sharing the centrally attended color and RDKs sharing the centrally ignored color. Hence, alpha-band modulations seem not to index the sustained global selection of attended over unattended feature values within the same feature dimension.

Sleep-like unsupervised replay reduces catastrophic forgetting in artificial neural networks

Artificial neural networks are known to suffer from catastrophic forgetting: when learning multiple tasks sequentially, they perform well on the most recent task at the expense of previously learned tasks. In the brain, sleep is known to play an important role in incremental learning by replaying recent and old conflicting memory traces. Here we tested the hypothesis that implementing a sleep-like phase in artificial neural networks can protect old memories during new training and alleviate catastrophic forgetting. Sleep was implemented as off-line training with local unsupervised Hebbian plasticity rules and noisy input. In an incremental learning framework, sleep was able to recover old tasks that were otherwise forgotten. Previously learned memories were replayed spontaneously during sleep, forming unique representations for each class of inputs. Representational sparseness and neuronal activity corresponding to the old tasks increased while new task related activity decreased. The study suggests that spontaneous replay simulating sleep-like dynamics can alleviate catastrophic forgetting in artificial neural networks.

Spatial attention in perceptual decision making as revealed by response-locked classification image analysis

In many situations, humans serially sample information from many locations in an image to make an appropriate decision about a visual target. Spatial attention and eye movements play a crucial role in this serial vision process. To investigate the effect of spatial attention in such dynamic decision making, we applied a classification image (CI) analysis locked to the observer's reaction time (RT). We asked human observers to detect as rapidly as possible a target whose contrast gradually increased on the left or right side of dynamic noise, with the presentation of a spatial cue. The analysis revealed a spatiotemporally biphasic profile of the CI which peaked at ~ 350 ms before the observer's response. We found that a valid cue presented at the target location shortened the RT and increased the overall amplitude of the CI, especially when the cue appeared 500-1250 ms before the observer's response. The results were quantitatively accounted for by a simple perceptual decision mechanism that accumulates the outputs of the spatiotemporal contrast detector, whose gain is increased by sustained attention to the cued location.

Disinhibitory circuit mediated by connections from vasoactive intestinal polypeptide to somatostatin interneurons underlies the paradoxical decrease in spike synchrony with increased border ownership selective neuron firing rate

The activity of border ownership selective (BOS) neurons in intermediate-level visual areas indicates which side of a contour owns a border relative to its classical receptive field and provides a fundamental component of figure-ground segregation. A physiological study reported that selective attention facilitates the activity of BOS neurons with a consistent border ownership preference, defined as two neurons tuned to respond to the same visual object. However, spike synchrony between this pair is significantly suppressed by selective attention. These neurophysiological findings are derived from a biologically-plausible microcircuit model consisting of spiking neurons including two subtypes of inhibitory interneurons, somatostatin (SOM) and vasoactive intestinal polypeptide (VIP) interneurons, and excitatory BOS model neurons. In our proposed model, BOS neurons and SOM interneurons cooperate and interact with each other. VIP interneurons not only suppress SOM interneuron responses but also are activated by feedback signals mediating selective attention, which leads to disinhibition of BOS neurons when they are directing selective attention toward an object. Our results suggest that disinhibition arising from the synaptic connections from VIP to SOM interneurons plays a critical role in attentional modulation of neurons in intermediate-level visual areas.

Hierarchical Bayesian LSTM for Head Trajectory Prediction on Omnidirectional Images

When viewing omnidirectional images (ODIs), viewers can access different viewports via head movement (HM), which sequentially forms head trajectories in spatial-temporal domain. Thus, head trajectories play a key role in modeling human attention on ODIs. In this paper, we establish a large-scale dataset collecting 21,600 head trajectories on 1,080 ODIs. By mining our dataset, we find two important factors influencing head trajectories, i.e., temporal dependency and subject-specific variance. Accordingly, we propose a novel approach integrating hierarchical Bayesian inference into long short-term memory (LSTM) network for head trajectory prediction on ODIs, which is called HiBayes-LSTM. In HiBayes-LSTM, we develop a mechanism of Future Intention Estimation (FIE), which captures the temporal correlations from previous, current and estimated future information, for predicting viewport transition. Additionally, a training scheme called Hierarchical Bayesian inference (HBI) is developed for modeling inter-subject uncertainty in HiBayes-LSTM. For HBI, we introduce a joint Gaussian distribution in a hierarchy, to approximate the posterior distribution over network weights. By sampling subject-specific weights from the approximated posterior distribution, our HiBayes-LSTM approach can yield diverse viewport transition among different subjects and obtain multiple head trajectories. Extensive experiments validate that our HiBayes-LSTM approach significantly outperforms 9 state-of-the-art approaches for trajectory prediction on ODIs, and then it is successfully applied to predict saliency on ODIs.

Analysis based on neural representation of natural object surfaces to elucidate the mechanisms of a trained AlexNet model

Analysis and understanding of trained deep neural networks (DNNs) can deepen our understanding of the visual mechanisms involved in primate visual perception. However, due to the limited availability of neural activity data recorded from various cortical areas, the correspondence between the characteristics of artificial and biological neural responses for visually recognizing objects remains unclear at the layer level of DNNs. In the current study, we investigated the relationships between the artificial representations in each layer of a trained AlexNet model (based on a DNN) for object classification and the neural representations in various levels of visual cortices such as the primary visual (V1), intermediate visual (V4), and inferior temporal cortices. Furthermore, we analyzed the profiles of the artificial representations at a single channel level for each layer of the AlexNet model. We found that the artificial representations in the lower-level layers of the trained AlexNet model were strongly correlated with the neural representation in V1, whereas the responses of model neurons in layers at the intermediate and higher-intermediate levels of the trained object classification model exhibited characteristics similar to those of neural activity in V4 neurons. These results suggest that the trained AlexNet model may gradually establish artificial representations for object classification through the hierarchy of its network, in a similar manner to the neural mechanisms by which afferent transmission beginning in the low-level features gradually establishes object recognition as signals progress through the hierarchy of the ventral visual pathway.

Weakly Supervised Visual Saliency Prediction

The success of current deep saliency models heavily depends on large amounts of annotated human fixation data to fit the highly non-linear mapping between the stimuli and visual saliency. Such fully supervised data-driven approaches are annotation-intensive and often fail to consider the underlying mechanisms of visual attention. In contrast, in this paper, we introduce a model based on various cognitive theories of visual saliency, which learns visual attention patterns in a weakly supervised manner. Our approach incorporates insights from cognitive science as differentiable submodules, resulting in a unified, end-to-end trainable framework. Specifically, our model encapsulates the following important components motivated from biological vision. (a) As scene semantics are closely related to visually attentive regions, our model encodes discriminative spatial information for scene understanding through spatial visual semantics embedding. (b) To model the objectness factors in visual attention deployment, we incorporate object-level semantics embedding and object relation information. (c) Considering the "winner-take-all" mechanism in visual stimuli processing, we model the competition mechanism among objects with softmax based neural attention. (d) Lastly, a conditional center prior is learned to mimic the spatial distribution bias of visual attention. Furthermore, we propose novel loss functions to utilize supervision cues from image-level semantics, saliency prior knowledge, and self-information compression. Experiments show that our method achieves promising results, and even outperforms many of its fully supervised counterparts. Overall, our weakly supervised saliency method makes an essential step towards reducing the annotation budget of current approaches, as well as providing a more comprehensive understanding of the visual attention mechanism. Our code is available at: https://github.com/ashleylqx/WeakFixation.git.

The role of temporal cortex in the control of attention

Attention is an indispensable component of active vision. Contrary to the widely accepted notion that temporal cortex processing primarily focusses on passive object recognition, a series of very recent studies emphasize the role of temporal cortex structures, specifically the superior temporal sulcus (STS) and inferotemporal (IT) cortex, in guiding attention and implementing cognitive programs relevant for behavioral tasks. The goal of this theoretical paper is to advance the hypothesis that the temporal cortex attention network (TAN) entails necessary components to actively participate in attentional control in a flexible task-dependent manner. First, we will briefly discuss the general architecture of the temporal cortex with a focus on the STS and IT cortex of monkeys and their modulation with attention. Then we will review evidence from behavioral and neurophysiological studies that support their guidance of attention in the presence of cognitive control signals. Next, we propose a mechanistic framework for executive control of attention in the temporal cortex. Finally, we summarize the role of temporal cortex in implementing cognitive programs and discuss how they contribute to the dynamic nature of visual attention to ensure flexible behavior.

Explaining Orientation Adaptation in V1 by Updating the State of a Spatial Model

In this work, we extend an influential statistical model based on the spatial classical receptive field (CRF) and non-classical receptive field (nCRF) interactions (Coen-Cagli et al., 2012) to explain the typical orientation adaptation effects observed in V1. If we assume that the temporal adaptation modifies the “state” of the model, the spatial statistical model can explain all of the orientation adaptation effects in the context of neuronal output using small and large grating observed in neurophysiological experiments in V1. The “state” of the model represents the internal parameters such as the prior and the covariance trained on a mixed dataset that totally determine the response of the model. These two parameters, respectively, reflect the probability of the orientation component and the connectivity among neurons between CRF and nCRF. Specifically, we have two key findings: First, neural adapted results using a small grating that just covers the CRF can be predicted by the change of the prior of our model. Second, the change of the prior can also predict most of the observed results using a large grating that covers both CRF and nCRF of a neuron. However, the prediction of the novel attractive adaptation using large grating covering both CRF and nCRF also necessitates the involvement of a connectivity change of the center-surround RFs. In addition, our paper contributes a new prior-based winner-take-all (WTA) working mechanism derived from the statistical-based model to explain why and how all of these orientation adaptation effects can be predicted by relying on this spatial model without modifying its structure, a novel application of the spatial model. The research results show that adaptation may link time and space by changing the “state” of the neural system according to a specific adaptor. Furthermore, different forms of stimulus used for adaptation can cause various adaptation effects, such as an a priori shift or a connectivity change, depending on the specific stimulus size.

Suppression of Unattended Features Is Independent of Task Relevance

Feature-based attention serves the separation of relevant from irrelevant features. While global amplification of attended features is coherently described as a key mechanism for feature-based attention, nature and constituting factors of neural suppressive interactions are far less clear. One aspect of global amplification is its flexible modulation by the task relevance of the to-be-attended stimulus. We examined whether suppression is similarly modulated by their respective task relevance or is mandatory for all unattended features. For this purpose, participants saw a display of randomly moving dots with 3 distinct colors and were asked to report brief events of coherent motion for a cued color. Of the 2 unattended colored clouds, one contained distracting motion events while the other was irrelevant and without such motion events throughout the experiment. We used electroencephalography-derived steady-state visual-evoked potentials to investigate early visual processing of the attended, unattended, and irrelevant color under sustained feature-based attention. The analysis revealed a biphasic process with an early amplification of the to-be-attended color followed by suppression of the to-be-ignored color relative to a pre-cue baseline. Importantly, the neural dynamics for the unattended and always irrelevant color were comparable. Suppression is thus a mandatory mechanism affecting all unattended stimuli irrespective of their task relevance.

Time-dependent inhibition of covert shifts of attention

Visual transients can interrupt overt orienting by abolishing the execution of a planned eye movement due about 90 ms later, a phenomenon known as saccadic inhibition (SI). It is not known if the same inhibitory process might influence covert orienting in the absence of saccades, and consequently alter visual perception. In Experiment 1 (n = 14), we measured orientation discrimination during a covert orienting task in which an uninformative exogenous visual cue preceded the onset of an oriented probe by 140-290 ms. In half of the trials, the onset of the probe was accompanied by a brief irrelevant flash, a visual transient that would normally induce SI. We report a time-dependent inhibition of covert orienting in which the irrelevant flash impaired orientation discrimination accuracy when the probe followed the cue by 190 and 240 ms. The interference was more pronounced when the cue was incongruent with the probe location, suggesting an impact on the reorienting component of the attentional shift. In Experiment 2 (n = 12), we tested whether the inhibitory effect of the flash could occur within an earlier time range, or only within the later, reorienting range. We presented probes at congruent cue locations in a time window between 50 and 200 ms. Similar to Experiment 1, discrimination performance was altered at 200 ms after the cue. We suggest that covert attention may be susceptible to similar inhibitory mechanisms that generate SI, especially in later stages of attentional shifting (> 200 ms after a cue), typically associated with reorienting.

Revisiting Persistent Neuronal Activity During Covert Spatial Attention

Persistent activity has been observed in the prefrontal cortex (PFC), in particular during the delay periods of visual attention tasks. Classical approaches based on the average activity over multiple trials have revealed that such an activity encodes the information about the attentional instruction provided in such tasks. However, single-trial approaches have shown that activity in this area is rather sparse than persistent and highly heterogeneous not only within the trials but also between the different trials. Thus, this observation raised the question of how persistent the actually persistent attention-related prefrontal activity is and how it contributes to spatial attention. In this paper, we review recent evidence of precisely deconstructing the persistence of the neural activity in the PFC in the context of attention orienting. The inclusion of machine-learning methods for decoding the information reveals that attention orienting is a highly dynamic process, possessing intrinsic oscillatory dynamics working at multiple timescales spanning from milliseconds to minutes. Dimensionality reduction methods further show that this persistent activity dynamically incorporates multiple sources of information. This novel framework reflects a high complexity in the neural representation of the attention-related information in the PFC, and how its computational organization predicts behavior.

Dynamics of a Mutual Inhibition Circuit between Pyramidal Neurons Compared to Human Perceptual Competition

Neural competition plays an essential role in active selection processes of noisy and ambiguous input signals, and it is assumed to underlie emergent properties of brain functioning, such as perceptual organization and decision-making. Despite ample theoretical research on neural competition, experimental tools to allow neurophysiological investigation of competing neurons have not been available. We developed a "hybrid" system where real-life neurons and a computer-simulated neural circuit interacted. It enabled us to construct a mutual inhibition circuit between two real-life pyramidal neurons. We then asked what dynamics this minimal unit of neural competition exhibits and compared them with the known behavioral-level dynamics of neural competition. We found that the pair of neurons shows bistability when activated simultaneously by current injections. The addition of modeled synaptic noise and changes in the activation strength showed that the dynamics of the circuit are strikingly similar to the known properties of bistable visual perception: The distribution of dominance durations showed a right-skewed shape, and the changes of the activation strengths caused changes in dominance, dominance durations, and reversal rates as stated in the well-known empirical laws of bistable perception known as Levelt's propositions.SIGNIFICANCE STATEMENT Visual perception emerges as the result of neural systems actively organizing visual signals that involves selection processes of competing neurons. While the neural competition, realized by a "mutual inhibition" circuit has been examined in many theoretical studies, its properties have not been investigated in real neurons. We have developed a "hybrid" system where two real-life pyramidal neurons in a mouse brain slice interact through a computer-simulated mutual inhibition circuit. We found that simultaneous activation of the neurons leads to bistable activity. We investigated the effect of noise and the effect of changes in the activation strength on the dynamics. We observed that the pair of neurons exhibit dynamics strikingly similar to the known properties of bistable visual perception.

Correspondence between Monkey Visual Cortices and Layers of a Saliency Map Model Based on a Deep Convolutional Neural Network for Representations of Natural Images

Attentional selection is a function that allocates the brain’s computational resources to the most important part of a visual scene at a specific moment. Saliency map models have been proposed as computational models to predict attentional selection within a spatial location. Recent saliency map models based on deep convolutional neural networks (DCNNs) exhibit the highest performance for predicting the location of attentional selection and human gaze, which reflect overt attention. Trained DCNNs potentially provide insight into the perceptual mechanisms of biological visual systems. However, the relationship between artificial and neural representations used for determining attentional selection and gaze location remains unknown. To understand the mechanism underlying saliency map models based on DCNNs and the neural system of attentional selection, we investigated the correspondence between layers of a DCNN saliency map model and monkey visual areas for natural image representations. We compared the characteristics of the responses in each layer of the model with those of the neural representation in the primary visual (V1), intermediate visual (V4), and inferior temporal (IT) cortices. Regardless of the DCNN layer level, the characteristics of the responses were consistent with that of the neural representation in V1. We found marked peaks of correspondence between V1 and the early level and higher-intermediate-level layers of the model. These results provide insight into the mechanism of the trained DCNN saliency map model and suggest that the neural representations in V1 play an important role in computing the saliency that mediates attentional selection, which supports the V1 saliency hypothesis.

The human endogenous attentional control network includes a ventro-temporal cortical node

Endogenous attention is the cognitive function that selects the relevant pieces of sensory information to achieve goals and it is known to be controlled by dorsal fronto-parietal brain areas. Here we expand this notion by identifying a control attention area located in the temporal lobe. By combining a demanding behavioral paradigm with functional neuroimaging and diffusion tractography, we show that like fronto-parietal attentional areas, the human posterior inferotemporal cortex exhibits significant attentional modulatory activity. This area is functionally distinct from surrounding cortical areas, and is directly connected to parietal and frontal attentional regions. These results show that attentional control spans three cortical lobes and overarches large distances through fiber pathways that run orthogonally to the dominant anterior-posterior axes of sensory processing, thus suggesting a different organizing principle for cognitive control.

Neural circuitry for stimulus selection in the zebrafish visual system

When navigating the environment, animals need to prioritize responses to the most relevant stimuli. Although a theoretical framework for selective visual attention exists, its circuit implementation has remained obscure. Here we investigated how larval zebrafish select between simultaneously presented visual stimuli. We found that a mix of winner-take-all (WTA) and averaging strategies best simulates behavioral responses. We identified two circuits whose activity patterns predict the relative saliencies of competing visual objects. Stimuli presented to only one eye are selected by WTA computation in the inner retina. Binocularly presented stimuli, on the other hand, are processed by reciprocal, bilateral connections between the nucleus isthmi (NI) and the tectum. This interhemispheric computation leads to WTA or averaging responses. Optogenetic stimulation and laser ablation of NI neurons disrupt stimulus selection and behavioral action selection. Thus, depending on the relative locations of competing stimuli, a combination of retinotectal and isthmotectal circuits enables selective visual attention.

Differential impact of exogenous and endogenous attention on the contrast sensitivity function across eccentricity

Both exogenous and endogenous covert spatial attention enhance contrast sensitivity, a fundamental measure of visual function that depends substantially on the spatial frequency and eccentricity of a stimulus. Whether and how each type of attention systematically improves contrast sensitivity across spatial frequency and eccentricity are fundamental to our understanding of visual perception. Previous studies have assessed the effects of spatial attention at individual spatial frequencies and, separately, at different eccentricities, but this is the first study to do so parametrically with the same task and observers. Using an orientation discrimination task, we investigated the effect of attention on contrast sensitivity over a wide range of spatial frequencies and eccentricities. Targets were presented alone or among distractors to assess signal enhancement and distractor suppression mechanisms of spatial attention. At each eccentricity, we found that exogenous attention preferentially enhanced spatial frequencies higher than the peak frequency in the baseline condition. In contrast, endogenous attention similarly enhanced a broad range of lower and higher spatial frequencies. The presence or absence of distractors did not alter the pattern of enhancement by each type of attention. Our findings reveal how the two types of covert spatial attention differentially shape how we perceive basic visual dimensions across the visual field.

The neural bases of spatial attention and perceptual rhythms

Attentional processes allow the brain to overcome its processing capacities limitations by enhancing relevant visual information and suppressing irrelevant information. Thus attention plays a critical role, shaping our perception of the world. Several models have been proposed to describe the neuronal bases of attention and its mechanistic underlyings. Recent electrophysiological evidence show that attentional processes rely on oscillatory brain activities that correlate with rhythmic changes in cognitive performance. In the present review, we first take a historical perspective on how attention is viewed, from the initial spotlight theory of attention to the recent dynamic view of attention selection and we review their supporting psychophysical evidence. Based on recent prefrontal electrophysiological evidence, we refine the most recent models of attention sampling by proposing a rhythmic and continuous model of attentional sampling. In particular, we show that attention involves a continuous exploration of space, shifting within and across visual hemifield at specific alpha and theta rhythms, independently of the current attentional load. In addition, we show that this prefrontal attentional spotlight implements conjointly selection and suppression mechanisms, and is captured by salient incoming items. Last, we argue that this attention spotlight implements a highly flexible alternation of attentional exploration and exploitation epochs, depending on ongoing task contingencies. In a last part, we review the local and network oscillatory mechanisms that correlate with rhythmic attentional sampling, describing multiple rhythmic generators and complex network interactions.

An Attempt at a Unified Theory of the Neocortical Microcircuit in Sensory Cortex

The neocortex performs a wide range of functions, including working memory, sensory perception, and motor planning. Despite this diversity in function, evidence suggests that the neocortex is made up of repeating subunits ("macrocolumns"), each of which is largely identical in circuitry. As such, the specific computations performed by these macrocolumns are of great interest to neuroscientists and AI researchers. Leading theories of this microcircuit include models of predictive coding, hierarchical temporal memory (HTM), and Adaptive Resonance Theory (ART). However, these models have not yet explained: (1) how microcircuits learn sequences input with delay (i.e., working memory); (2) how networks of columns coordinate processing on precise timescales; or (3) how top-down attention modulates sensory processing. I provide a theory of the neocortical microcircuit that extends prior models in all three ways. Additionally, this theory provides a novel working memory circuit that extends prior models to support simultaneous multi-item storage without disrupting ongoing sensory processing. I then use this theory to explain the functional origin of a diverse set of experimental findings, such as cortical oscillations.

Sex differences in implicit processing of allocentric relationships between objects and location in a Simon task

Simon tasks reveal implicit processing conflicts that arise when the abstract coding of stimulus position is incongruent with coding for location of the output response. Participants were tested with two versions of a Simon task in a counterbalanced order to examine a potential female bias for attending to object characteristics versus object location. Both tasks used a triangle pointing to the left or right. A simple version presented the triangle in an inner or outer position relative to central fixation. A more complex version included a frame surrounding the inner-outer triangle presentation area in order to introduce additional visual elements for left/right visual processing. When the No Frame version was the first presented, there were no sex differences in the Simon effect in either version, which is consistent with results from other studies that did not provide feedback regarding accuracy. When the initial test was the Frame version, we observed a reverse Simon effect for incongruent triangles presented in the left inner position, with females faster than males to identify the incongruent condition versus the congruent (-59 vs -5 msec). In the No Frame condition that followed, females showed a carryover effect from the previous Frame condition, exhibiting positive Simon effects that were two fold larger than males for identifying incongruent stimuli presented in the left and right outer positions. Similar to previous Simon studies, females showed longer overall reaction times than males (~15%). The difference was not related to the Simon effect and is also found in other types of tasks involving early visual processing of objects with location. Based on sex differences in the Simon effect that emerged following initial experience of the triangle adjoining the frame, the present results support a female bias toward broader integration of objects within the context of location.

The coordinate system of endogenous spatial attention during smooth pursuit

A central question in vision is whether spatial attention is represented in an eye-centered (retinotopic) or world-centered (spatiotopic) reference-frame. Most previous studies on this question focused on how coordinates are modulated across saccades. In the present study, we investigated the reference-frame of attention across smooth pursuit eye-movements using a goal-directed saccade task. In two experiments, participants were asked to pursue a moving target while attending to one or two grating stimuli. On each trial, one stimulus was constant in its retinal position and the other was constant in its spatial position. Upon detection of a slight change in stimulus orientation, participants were asked to stop pursuing and perform a fast saccade toward the modified stimulus. In the focused attention condition, they attended one, predefined, stimulus, and in the divided attention condition they attended both. In Experiment 1 the angle of the orientation change marking the target event was constant across participants and conditions. In Experiment 2, the angle was individually adapted to equate performance across participants and conditions. Findings of the two experiments were consistent and showed that the enhancement of mean visual sensitivity in the focused relative to the divided attention condition was similar in magnitude for both retinotopic and spatiotopic targets. This indicates that during smooth pursuit, endogenous attention was proportionally divided between targets in retinotopic and spatiotopic frames of reference.

Competing rhythmic neural representations of orientations during concurrent attention to multiple orientation features

When a feature is attended, all locations containing this feature are enhanced throughout the visual field. However, how the brain concurrently attends to multiple features remains unknown and cannot be easily deduced from classical attention theories. Here, we recorded human magnetoencephalography signals when subjects concurrently attended to two spatially overlapping orientations. A time-resolved multivariate inverted encoding model was employed to track the ongoing temporal courses of the neural representations of the attended orientations. We show that the two orientation representations alternate with each other and undergo a theta-band (~4 Hz) rhythmic fluctuation over time. Similar temporal profiles are also revealed in the orientation discrimination performance. Computational modeling suggests a tuning competition process between the two neuronal populations that are selectively tuned to one of the attended orientations. Taken together, our findings reveal for the first time a rhythm-based, time-multiplexing neural machinery underlying concurrent multi-feature attention.

The neural mechanisms for concurrently attending to multiple features in the visual stimuli are not well understood. Here, the authors show that the neural representations for two overlapping stimulus features alternate with each other at a ~4 Hz rhythm that was also observed in fluctuations in the task performance.

Evidence for an attentional priority map in inferotemporal cortex

A fundamental dogma in the cognitive neurosciences is that attention is controlled by parietal and prefrontal areas. Here, we show that an area in the temporal lobe exhibits the properties of a priority map coding the focus of attention. We show this through whole-brain functional magnetic resonance imaging, electrophysiological single-unit recordings, and causal electrical stimulation. This discovery changes our understanding of the organization of visual pathways and the functions of attention networks.

From incoming sensory information, our brains make selections according to current behavioral goals. This process, selective attention, is controlled by parietal and frontal areas. Here, we show that another brain area, posterior inferotemporal cortex (PITd), also exhibits the defining properties of attentional control. We discovered this area with functional magnetic resonance imaging (fMRI) during an attentive motion discrimination task. Single-cell recordings from PITd revealed strong attentional modulation across 3 attention tasks yet no tuning to task-relevant stimulus features, like motion direction or color. Instead, PITd neurons closely tracked the subject’s attention state and predicted upcoming errors of attentional selection. Furthermore, artificial electrical PITd stimulation controlled the location of attentional selection without altering feature discrimination. These are the defining properties of a feature-blind priority map encoding the locus of attention. Together, these results suggest area PITd, located strategically to gather information about object properties, as an attentional priority map.

Salience Models: A Computational Cognitive Neuroscience Review

The seminal model by Laurent Itti and Cristoph Koch demonstrated that we can compute the entire flow of visual processing from input to resulting fixations. Despite many replications and follow-ups, few have matched the impact of the original model-so what made this model so groundbreaking? We have selected five key contributions that distinguish the original salience model by Itti and Koch; namely, its contribution to our theoretical, neural, and computational understanding of visual processing, as well as the spatial and temporal predictions for fixation distributions. During the last 20 years, advances in the field have brought up various techniques and approaches to salience modelling, many of which tried to improve or add to the initial Itti and Koch model. One of the most recent trends has been to adopt the computational power of deep learning neural networks; however, this has also shifted their primary focus to spatial classification. We present a review of recent approaches to modelling salience, starting from direct variations of the Itti and Koch salience model to sophisticated deep-learning architectures, and discuss the models from the point of view of their contribution to computational cognitive neuroscience.

Spike-Based Winner-Take-All Computation: Fundamental Limits and Order-Optimal Circuits

Winner-take-all (WTA) refers to the neural operation that selects a (typically small) group of neurons from a large neuron pool. It is conjectured to underlie many of the brain's fundamental computational abilities. However, not much is known about the robustness of a spike-based WTA network to the inherent randomness of the input spike trains. In this work, we consider a spike-based k-WTA model wherein n randomly generated input spike trains compete with each other based on their underlying firing rates and k winners are supposed to be selected. We slot the time evenly with each time slot of length 1 ms and model the n input spike trains as n independent Bernoulli processes. We analytically characterize the minimum waiting time needed so that a target minimax decision accuracy (success probability) can be reached. We first derive an information-theoretic lower bound on the waiting time. We show that to guarantee a (minimax) decision error ≤δ (where δ∈(0,1)), the waiting time of any WTA circuit is at least [Formula: see text]where R⊆(0,1) is a finite set of rates and TR is a difficulty parameter of a WTA task with respect to set R for independent input spike trains. Additionally, TR is independent of δ, n, and k. We then design a simple WTA circuit whose waiting time is [Formula: see text]provided that the local memory of each output neuron is sufficiently long. It turns out that for any fixed δ, this decision time is order-optimal (i.e., it matches the above lower bound up to a multiplicative constant factor) in terms of its scaling in n, k, and TR.

Distributed attention model of perceptual averaging

The visual system efficiently processes complex and redundant information in a scene despite its limited capacity. One strategy for coping with the complexity and redundancy of a scene is to summarize it by using average information. However, despite its importance, the mechanism of averaging is not well understood. Here, a distributed attention model of averaging is proposed. Human percept for an object can be disturbed by various sources of internal noise, which can occur either before (early noise) or after (late noise) forming an ensemble perception. The model assumes these noises and reflects noise cancellation by averaging multiple items. The model predicts increased precision for more items with decelerated increments for large set-sizes resulting from late noise. Importantly, the model incorporates mechanisms of attention, which modulate each item's contribution to the averaging process. The attention in the model also results in saturation of performance increments for small set-sizes because the amount of attention allocated to each item is greater for small set-sizes than for large set-sizes. To evaluate the proposed model, a psychophysical experiment was conducted in which observers' ability to discriminate average sizes of two displays was measured. The observers' averaging performance increased at a decreasing rate with small set-sizes and it approached an asymptote for large set-sizes. The model accurately predicted the observed pattern of data. It provides a theoretical framework for interpreting behavioral data and leads to an understanding of the characteristics of ensemble perception.

Neural activity of heterogeneous inhibitory spiking networks with delay

We study a network of spiking neurons with heterogeneous excitabilities connected via inhibitory delayed pulses. For globally coupled systems the increase of the inhibitory coupling reduces the number of firing neurons by following a winner-takes-all mechanism. For sufficiently large transmission delay we observe the emergence of collective oscillations in the system beyond a critical coupling value. Heterogeneity promotes neural inactivation and asynchronous dynamics and its effect can be counteracted by considering longer time delays. In sparse networks, inhibition has the counterintuitive effect of promoting neural reactivation of silent neurons for sufficiently large coupling. In this regime, current fluctuations are on one side responsible for neural firing of subthreshold neurons and on the other side for their desynchronization. Therefore, collective oscillations are present only in a limited range of coupling values, which remains finite in the thermodynamic limit. Out of this range the dynamics is asynchronous and for very large inhibition neurons display a bursting behavior alternating periods of silence with periods where they fire freely in absence of any inhibition.

Variation in target and distractor heterogeneity impacts performance in the centroid task

In a selective centroid task, the participant views a brief cloud of items of different types-some of which are targets, the others distractors-and strives to mouse-click the centroid of the target items, ignoring the distractors. Advantages of the centroid task are that multiple target types can appear in the same display and that influence functions, which estimate the weight of each stimulus type in the cloud on the perceived centroid for each participant, can be obtained easily and efficiently. Here we document the strong, negative impact on performance that results when the participant is instructed to attend to target dots that consist of two or more levels of a single feature dimension, even when those levels differ categorically from those of the distractor dots. The results also show a smaller, but still observable decrement in performance that results when there is heterogeneity in the distractor dots.

Optimality and heuristics in perceptual neuroscience

The foundation for modern understanding of how we make perceptual decisions about what we see or where to look comes from considering the optimal way to perform these behaviors. While statistical computation is useful for deriving the optimal solution to a perceptual problem, optimality requires perfect knowledge of priors and often complex computation. Accumulating evidence, however, suggests that optimal perceptual goals can be achieved or approximated more simply by human observers using heuristic approaches. Perceptual neuroscientists captivated by optimal explanations of sensory behaviors will fail in their search for the neural circuits and cortical processes that implement an optimal computation whenever that behavior is actually achieved through heuristics. This article provides a cross-disciplinary review of decision-making with the aim of building perceptual theory that uses optimality to set the computational goals for perceptual behavior but, through consideration of ecological, computational, and energetic constraints, incorporates how these optimal goals can be achieved through heuristic approximation.

The attentional template is shifted and asymmetrically sharpened by distractor context

Theories of attention hypothesize the existence of an "attentional template" that contains target features in working or long-term memory. It is often assumed that the template contents are veridical, but recent studies have found that this is not true when the distractor set is linearly separable from the target (e.g., all distractors are "yellower" than an orange-colored target). In such cases, the target representation in memory shifts away from distractor features (Navalpakkam & Itti, 2007) and develops a sharper boundary with distractors (Geng, DiQuattro, & Helm, 2017). These changes in the target template are presumed to increase the target-to-distractor psychological distinctiveness and lead to better attentional selection, but it remains unclear what characteristics of the distractor context produce shifting versus sharpening. Here, we tested the hypothesis that the template representation shifts whenever the distractor set (i.e., all of the distractors) is linearly separable from the target but asymmetrical sharpening occurs only when linearly separable distractors are highly target-similar. Our results were consistent, suggesting that template shifting and asymmetrical sharpening are 2 mechanisms that increase the representational distinctiveness of targets from expected distractors and improve visual search performance. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Neuronal firing rates diverge during REM and homogenize during non-REM

Neurons fire at highly variable intrinsic rates and recent evidence suggests that low- and high-firing rate neurons display different plasticity and dynamics. Furthermore, recent publications imply possibly differing rate-dependent effects in hippocampus versus neocortex, but those analyses were carried out separately and with potentially important differences. To more effectively synthesize these questions, we analyzed the firing rate dynamics of populations of neurons in both hippocampal CA1 and frontal cortex under one framework that avoids the pitfalls of previous analyses and accounts for regression to the mean (RTM). We observed several consistent effects across these regions. While rapid eye movement (REM) sleep was marked by decreased hippocampal firing and increased neocortical firing, in both regions firing rate distributions widened during REM due to differential changes in high- versus low-firing rate cells in parallel with increased interneuron activity. In contrast, upon non-REM (NREM) sleep, firing rate distributions narrowed while interneuron firing decreased. Interestingly, hippocampal interneuron activity closely followed the patterns observed in neocortical principal cells rather than the hippocampal principal cells, suggestive of long-range interactions. Following these undulations in variance, the net effect of sleep was a decrease in firing rates. These decreases were greater in lower-firing hippocampal neurons but also higher-firing frontal cortical neurons, suggestive of greater plasticity in these cell groups. Our results across two different regions, and with statistical corrections, indicate that the hippocampus and neocortex show a mixture of differences and similarities as they cycle between sleep states with a unifying characteristic of homogenization of firing during NREM and diversification during REM.

Decisions are expedited through multiple neural adjustments spanning the sensorimotor hierarchy

When decisions are made under speed pressure, "urgency" signals elevate neural activity toward action-triggering thresholds independent of the sensory evidence, thus incurring a cost to choice accuracy. While urgency signals have been observed in brain circuits involved in preparing actions, their influence at other levels of the sensorimotor pathway remains unknown. We used a novel contrast-comparison paradigm to simultaneously trace the dynamics of sensory evidence encoding, evidence accumulation, motor preparation, and muscle activation in humans. Results indicate speed pressure impacts multiple sensorimotor levels but in crucially distinct ways. Evidence-independent urgency was applied to cortical action-preparation signals and downstream muscle activation, but not directly to upstream levels. Instead, differential sensory evidence encoding was enhanced in a way that partially countered the negative impact of motor-level urgency on accuracy, and these opposing sensory-boost and motor-urgency effects had knock-on effects on the buildup and pre-response amplitude of a motor-independent representation of cumulative evidence.

A Winner-Take-All Strategy for Improved Object Tracking

Recently, numerous state-of-the-art learning schemes are proposed for object tracking. However, typically, most methods can only solve certain type of challenges but are less effective for the rest-no single tracker is perfect for all challenges. In this paper, a winner-take-all (WTA) strategy is exploited to select a winner tracker (considering both accuracy and efficiency) from a set of prevailing methods to tackle the current challenge, according to features extracted from the current environment and an efficiency factor. To achieve this, a structural regression model to characterize the trackers is trained on a public dataset. By incorporating the complementary abilities from multiple trackers, the diversity of the model is improved so that the WTA tracker can tackle various unpredictable difficulties. Since only one tracker is selected at any time, the average efficiency of the proposed model is also higher than that of complex trackers in the tracker set. The proposed WTA framework is tested on two benchmark datasets as well as several long sequences, and extensive experimental results illustrate that WTA can significantly improve both the performance and the efficiency.