Spatial learning emerges not only from static environmental cues but also from the social and semantic context embedded in our surroundings. This study investigates how human agents influence visual exploration and spatial knowledge acquisition in a controlled Virtual Reality (VR) environment, focusing on the role of contextual congruency. Participants freely explored a 1 km2 virtual city while their eye movements were recorded. Agents were visually identical across conditions but placed in locations that were either congruent, incongruent, or neutral with respect to the surrounding environment. Using Bayesian hierarchical modeling, we found that incongruent agents elicited longer fixations and higher gaze transition entropy (GTE), a measure of scanning variability. Crucially, GTE emerged as the strongest predictor of spatial recall accuracy. A counterfactual mediation analysis indicated a small but reliable pathway via GTE and, for incongruent agents, a larger direct component not captured by GTE. These findings suggest that human-contextual incongruence promotes more flexible and distributed visual exploration, thereby enhancing spatial learning. By showing that human agents shape not only where we look but how we explore and encode space, this study contributes to a growing understanding of how social meaning guides attention and supports navigation.

The successful integration of Autonomous Vehicles (AVs) into daily life is influenced by various individual and contextual factors. This study examines key factors affecting acceptance and trust in AVs by assessing how interactive experiences influence user perceptions and adoption. Utilizing a Virtual Reality (VR) environment to simulate real-world driving conditions, the research investigates the impact of different levels of vehicle interaction—manual control, semi-autonomous, fully autonomous, and taxi-riding modes—on user and passenger trust, anxiety, perceived ease of use, and overall acceptance. A dual-method approach was employed, combining objective data collection through eye-tracking, which measures engagement and vigilance, with subjective user feedback gathered via the Autonomous Vehicle Acceptance Model (AVAM) questionnaire. The findings reveal a complex relationship between user control and trust in AV systems. While semi-autonomous conditions that incorporate user participation in decision-making enhance trust and intention to use, fully autonomous conditions influence trust but not intention to use significantly. However, semi- autonomous conditions also elicited higher anxiety levels compared to fully autonomous modes, suggesting a trade-off between interaction and comfort. Engagement levels, as indicated by visual metrics such as gaze patterns, were found to be critical for understanding user acceptance. These results highlight the importance of considering both subjective perceptions and objective behaviors in developing AV interfaces. The study contributes significant insights into the cognitive underpinnings of AV acceptance and underscores the need for strategies that both enhance user trust and minimize anxiety by finding an optimal balance between automation and user control. Further research into this balance is recommended to better accommodate the evolving nature of user trust and the cognitive complexities associated with semi-autonomous systems.

Most artificial neural networks used for object recognition are trained in a fully supervised setup. This is not only resource consuming as it requires large data sets of labeled examples but also quite different from how humans learn. We use a setup in which an artificial agent first learns in a simulated world through self-supervised, curiosity-driven exploration. Following this initial learning phase, the learned representations can be used to quickly associate semantic concepts such as different types of doors using one or more labeled examples. To do this, we use a method we call fast concept mapping which uses correlated firing patterns of neurons to define and detect semantic concepts. This association works instantaneously with very few labeled examples, similar to what we observe in humans in a phenomenon called fast mapping. Strikingly, we can already identify objects with as little as one labeled example which highlights the quality of the encoding learned self-supervised through interaction with the world. It therefore presents a feasible strategy for learning concepts without much supervision and shows that through pure interaction meaningful representations of an environment can be learned that work better for few-short learning than non-interactive methods.

Natural eye movements have primarily been studied for over-learned activities such as tea-making, sandwich-making, and hand-washing, which have a fixed sequence of associated actions. These studies demonstrate a sequential activation of low-level cognitive schemas facilitating task completion. However, whether these action schemas are activated in the same pattern when a task is novel and a sequence of actions must be planned in the moment is unclear. Here, we recorded gaze and body movements in a naturalistic task to study action-oriented gaze behavior. In a virtual environment, subjects moved objects on a life-size shelf to achieve a given order. To compel cognitive planning, we added complexity to the sorting tasks. Fixations aligned with the action onset showed gaze as tightly coupled with the action sequence, and task complexity moderately affected the proportion of fixations on the task-relevant regions. Our analysis revealed that gaze fixations were allocated to action-relevant targets just in time. Planning behavior predominantly corresponded to a greater visual search for task-relevant objects before the action onset. The results support the idea that natural behavior relies on the frugal use of working memory, and humans refrain from encoding objects in the environment to plan long-term actions. Instead, they prefer just-in-time planning by searching for action-relevant items at the moment, directing their body and hand to it, monitoring the action until it is terminated, and moving on to the following action.

In this work we apply a graph-theoretical analysis approach to eye tracking data recorded in virtual reality to investigate the underlying patterns of visual attention during spatial navigation. Based on the eye tracking data recorded in one virtual city, our graph-theoretical analysis identifies a subset of houses outstanding in their graph-theoretical properties which we define as gaze-graph-defined landmarks. Moreover, we are able to replicate these results with a different eye tracking data set recorded in a different virtual city. Finally, the initial model selection process of the participant’s performance in a point-to-building task in the second city suggests a stronger influence of graph-theoretical predictors on the performance compared to the non-graph related measures, however more research will be necessary to determine their relationship.

Language interfaces with many other cognitive domains. This paper explores how interactions at these interfaces can be studied with deep learning methods, focusing on the relation between language emergence and visual perception. To model the emergence of language, a sender and a receiver agent are trained on a reference game. The agents are implemented as deep neural networks, with dedicated vision and language modules. Motivated by the mutual influence between language and perception in cognition, we apply systematic manipulations to the agents' (i) visual representations, to analyze the effects on emergent communication, and (ii) communication protocols, to analyze the effects on visual representations. Our analyses show that perceptual biases shape semantic categorization and communicative content. Conversely, if the communication protocol partitions object space along certain attributes, agents learn to represent visual information about these attributes more accurately, and the representations of communication partners align. Finally, an evolutionary analysis suggests that visual representations may be shaped in part to facilitate the communication of environmentally relevant distinctions. Aside from accounting for co-adaptation effects between language and perception, our results point out ways to modulate and improve visual representation learning and emergent communication in artificial agents.

Vision provides the most important sensory information for spatial navigation. Recent technical advances allow new options to conduct more naturalistic experiments in virtual reality (VR) while additionally gathering data of the viewing behavior with eye tracking investigations. Here, we propose a method that allows one to quantify characteristics of visual behavior by using graph-theoretical measures to abstract eye tracking data recorded in a 3D virtual urban environment. The analysis is based on eye tracking data of 20 participants, who freely explored the virtual city Seahaven for 90 minutes with an immersive VR headset with an inbuild eye tracker. To extract what participants looked at, we defined "gaze" events, from which we created gaze graphs. On these, we applied graph-theoretical measures to reveal the underlying structure of visual attention. Applying graph partitioning, we found that our virtual environment could be treated as one coherent city. To investigate the importance of houses in the city, we applied the node degree centrality measure. Our results revealed that 10 houses had a node degree that exceeded consistently two-sigma distance from the mean node degree of all other houses. The importance of these houses was supported by the hierarchy index, which showed a clear hierarchical structure of the gaze graphs. As these high node degree houses fulfilled several characteristics of landmarks, we named them "gaze-graph-defined landmarks". Applying the rich club coefficient, we found that these gaze-graph-defined landmarks were preferentially connected to each other and that participants spend the majority of their experiment time in areas where at least two of those houses were visible. Our findings do not only provide new experimental evidence for the development of spatial knowledge, but also establish a new methodology to identify and assess the function of landmarks in spatial navigation based on eye tracking data.

Autonomous vehicles represent a significant development in our society, and their acceptance will largely depend on trust. This study investigates strategies to increase trust and acceptance by making the cars’ decisions. For this purpose, we created a virtual reality (VR) experiment with a self-explaining autonomous car, providing participants with verbal cues about crucial traffic decisions. First, we investigated attitudes toward self-driving cars among 7850 participants using a simplified version of the Technology Acceptance Model (TAM) questionnaire. Results revealed that female participants are less accepting than male participants, and that there is a general decline among all genders. Otherwise in general, a self-explaining car has a positive impact on trust and perceived usefulness. Surprisingly, it adversely affected the intention to use and perceived ease of use. This entails dissociation of trust from the other items of the questionnaire. Second, we analyzed behavioral of 26 750 participants to investigate the effect of self-explaining systems on head movements during the VR drive. We observed significant differences in head movements during the critical events and the baseline periods of the drive between the three driving conditions. Additionally, we demonstrated positive correlations between head movement parameters and the TAM scores, where trust showed the lowest correlation. This provides further evidence of the dissociation of trust from other TAM items. These findings illustrate the benefits of combining subjective questionnaire data with objective behavioral data. Overall, the outcomes indicate a partial dissociation of self-reported trust from intention to use and objective behavioral data.

Augmented reality applications allow users to enrich their real surroundings with additional digital content. However, due to the limited field of view of augmented reality devices, it can sometimes be difficult to become aware of newly emerging information inside or outside the field of view. Typical visual conflicts like clutter and occlusion of augmentations occur and can be further aggravated especially in the context of dense information spaces. In this article, we evaluate how multisensory cue combinations can improve the awareness for moving out-of-view objects in narrow field of view augmented reality displays. We distinguish between proximity and transition cues in either visual, auditory or tactile manner. Proximity cues are intended to enhance spatial awareness of approaching out-of-view objects while transition cues inform the user that the object just entered the field of view. In study 1, user preference was determined for 6 different cue combinations via forced-choice decisions. In study 2, the 3 most preferred modes were then evaluated with respect to performance and awareness measures in a divided attention reaction task. Both studies were conducted under varying noise levels. We show that on average the Visual-Tactile combination leads to 63% and Audio-Tactile to 65% faster reactions to incoming out-of-view augmentations than their Visual-Audio counterpart, indicating a high usefulness of tactile transition cues. We further show a detrimental effect of visual and audio noise on performance when feedback included visual proximity cues. Based on these results, we make recommendations to determine which cue combination is appropriate for which application.

How do humans acquire a meaningful understanding of the world with little to no supervision or semantic labels provided by the environment? Here we investigate embodiment with a closed loop between action and perception as one key component in this process. We take a close look at the representations learned by a deep reinforcement learning agent that is trained with high-dimensional visual observations collected in a 3D environment with very sparse rewards. We show that this agent learns stable representations of meaningful concepts such as doors without receiving any semantic labels. Our results show that the agent learns to represent the action relevant information, extracted from a simulated camera stream, in a wide variety of sparse activation patterns. The quality of the representations learned shows the strength of embodied learning and its advantages over fully supervised approaches.

Dynamic communication and routing play important roles in the human brain in order to facilitate flexibility in task solving and thought processes. Here, we present a network perturbation methodology that allows investigating dynamic switching between different network pathways based on phase offsets between two external oscillatory drivers. We apply this method in a computational model of the human connectome with delay-coupled neural masses. To analyze dynamic switching of pathways, we define four new metrics that measure dynamic network response properties for pairs of stimulated nodes. Evaluating these metrics for all network pathways, we found a broad spectrum of pathways with distinct dynamic properties and switching behaviors. We show that network pathways can have characteristic timescales and thus specific preferences for the phase lag between the regions they connect. Specifically, we identified pairs of network nodes whose connecting paths can either be (1) insensitive to the phase relationship between the node pair, (2) turned on and off via changes in the phase relationship between the node pair, or (3) switched between via changes in the phase relationship between the node pair. Regarding the latter, we found that 33% of node pairs can switch their communication from one pathway to another depending on their phase offsets. This reveals a potential mechanistic role that phase offsets and coupling delays might play for the dynamic information routing via communication pathways in the brain.

Current stress induction methods are often too theoretical and do not reflect to real life scenarios. In this study, we used a dual task paradigm in virtual reality, combining a navigation task with a reaction task. With this setup, we aimed at creating a novel benchmark stress induction approach utilizing modern virtual reality technology. Results show that our paradigm induced small scale physiological and subjective state changes. Lastly, we discuss our paradigm and experimental results from the perspective of ecological validity and present suggestions for improving our stress induction methodology as well as potential areas of use.

In this study, we investigate if phase-locking of fast oscillatory activity relies on the anatomical skeleton and if simple computational models informed by structural connectivity can help further to explain missing links in the structure-function relationship. We use diffusion tensor imaging data and alpha band-limited EEG signal recorded in a group of healthy individuals. Our results show that about 23.4% of the variance in empirical networks of resting-state functional connectivity is explained by the underlying white matter architecture. Simulating functional connectivity using a simple computational model based on the structural connectivity can increase the match to 45.4%. In a second step, we use our modeling framework to explore several technical alternatives along the modeling path. First, we find that an augmentation of homotopic connections in the structural connectivity matrix improves the link to functional connectivity while a correction for fiber distance slightly decreases the performance of the model. Second, a more complex computational model based on Kuramoto oscillators leads to a slight improvement of the model fit. Third, we show that the comparison of modeled and empirical functional connectivity at source level is much more specific for the underlying structural connectivity. However, different source reconstruction algorithms gave comparable results. Of note, as the fourth finding, the model fit was much better if zero-phase lag components were preserved in the empirical functional connectome, indicating a considerable amount of functionally relevant synchrony taking place with near zero or zero-phase lag. The combination of the best performing alternatives at each stage in the pipeline results in a model that explains 54.4% of the variance in the empirical EEG functional connectivity. Our study shows that large-scale brain circuits of fast neural network synchrony strongly rely upon the structural connectome and simple computational models of neural activity can explain missing links in the structure-function relationship.

The interest in saccadic IOR is funneled by the hypothesis that it serves a clear functional purpose in the selection of fixation points: the facilitation of foraging. In this study, we arrive at a different interpretation of saccadic IOR. First, we find that return saccades are performed much more often than expected from the statistical properties of saccades and saccade pairs. Second, we find that fixation durations before a saccade are modulated by the relative angle of the saccade, but return saccades show no sign of an additional temporal inhibition. Thus, we do not find temporal saccadic inhibition of return. Interestingly, we find that return locations are more salient, according to empirically measured saliency (locations that are fixated by many observers) as well as stimulus dependent saliency (defined by image features), than regular fixation locations. These results and the finding that return saccades increase the match of individual trajectories with a grand total priority map evidences the return saccades being part of a fixation selection strategy that trades off exploration and exploitation.

Visual attention is thought to be driven by the interplay between low-level visual features and task dependent information content of local image regions, as well as by spatial viewing biases. Though dependent on experimental paradigms and model assumptions, this idea has given rise to varying claims that either bottom-up or top-down mechanisms dominate visual attention. To contribute toward a resolution of this discussion, here we quantify the influence of these factors and their relative importance in a set of classification tasks. Our stimuli consist of individual image patches (bubbles). For each bubble we derive three measures: a measure of salience based on low-level stimulus features, a measure of salience based on the task dependent information content derived from our subjects' classification responses and a measure of salience based on spatial viewing biases. Furthermore, we measure the empirical salience of each bubble based on our subjects' measured eye gazes thus characterizing the overt visual attention each bubble receives. A multivariate linear model relates the three salience measures to overt visual attention. It reveals that all three salience measures contribute significantly. The effect of spatial viewing biases is highest and rather constant in different tasks. The contribution of task dependent information is a close runner-up. Specifically, in a standardized task of judging facial expressions it scores highly. The contribution of low-level features is, on average, somewhat lower. However, in a prototypical search task, without an available template, it makes a strong contribution on par with the two other measures. Finally, the contributions of the three factors are only slightly redundant, and the semi-partial correlation coefficients are only slightly lower than the coefficients for full correlations. These data provide evidence that all three measures make significant and independent contributions and that none can be neglected in a model of human overt visual attention.

Physiological experiments demonstrate the existence of weak pairwise correlations of neuronal activity in mammalian cortex (Singer, 1993). The functional implications of this correlated activity are hotly debated (Roskies et al., 1999). Nevertheless, it is generally considered a widespread feature of cortical dynamics. In recent years, another line of research has attracted great interest: the observation of a bimodal distribution of the membrane potential defining up states and down states at the single cell level (Wilson & Kawaguchi, 1996; Steriade, Contreras, & Amzica, 1994; Contreras & Steriade, 1995; Steriade, 2001). Here we use a theoretical approach to demonstrate that the latter phenomenon is a natural consequence of the former. In particular, we show that weak pairwise correlations of the inputs to a compartmental model of a layer V pyramidal cell can induce bimodality in its membrane potential. We show how this relationship can account for the observed increase of the power in the gamma-frequency band during up states, as well as the increase in the standard deviation and fraction of time spent in the depolarized state (Anderson, Lampl, Reichova, Carandini, & Ferster, 2000). In order to quantify the relationship between the correlation properties of a cortical network and the bistable dynamics of single neurons, we introduce a number of new indices. Subsequently, we demonstrate that a quantitative agreement with the experimental data can be achieved, introducing voltage-dependent mechanisms in our neuronal model such as Ca(2+)- and Ca(2+)-dependent K(+) channels. In addition, we show that the up states and down states of the membrane potential are dependent on the dendritic morphology of cortical neurons. Furthermore, bringing together network and single cell dynamics under a unified view allows the direct transfer of results obtained in one context to the other and suggests a new experimental paradigm: the use of specific intracellular analysis as a powerful tool to reveal the properties of the correlation structure present in the network dynamics.

Encoding of sensory events in internal states of the brain requires that this information can be decoded by other neural structures. The encoding of sensory events can involve both the spatial organization of neuronal activity and its temporal dynamics. Here we investigate the issue of decoding in the context of a recently proposed encoding scheme: the temporal population code. In this code, the geometric properties of visual stimuli become encoded into the temporal response characteristics of the summed activities of a population of cortical neurons. For its decoding, we evaluate a model based on the structure and dynamics of cortical microcircuits that is proposed for computations on continuous temporal streams: the liquid state machine. Employing the original proposal of the decoding network results in a moderate performance. Our analysis shows that the temporal mixing of subsequent stimuli results in a joint representation that compromises their classification. To overcome this problem, we investigate a number of initialization strategies. Whereas we observe that a deterministically initialized network results in the best performance, we find that in case the network is never reset, that is, it continuously processes the sequence of stimuli, the classification performance is greatly hampered by the mixing of information from past and present stimuli. We conclude that this problem of the mixing of temporally segregated information is not specific to this particular decoding model but relates to a general problem that any circuit that processes continuous streams of temporal information needs to solve. Furthermore, as both the encoding and decoding components of our network have been independently proposed as models of the cerebral cortex, our results suggest that the brain could solve the problem of temporal mixing by applying reset signals at stimulus onset, leading to a temporal segmentation of a continuous input stream.

Learning in neural networks is usually applied to parameters related to linear kernels and keeps the nonlinearity of the model fixed. Thus, for successful models, properties and parameters of the nonlinearity have to be specified using a priori knowledge, which often is missing. Here, we investigate adapting the nonlinearity simultaneously with the linear kernel. We use natural visual stimuli for training a simple model of the visual system. Many of the neurons converge to an energy detector matching existing models of complex cells. The overall distribution of the parameter describing the nonlinearity well matches recent physiological results. Controls with randomly shuffled natural stimuli and pink noise demonstrate that the match of simulation and experimental results depends on the higher-order statistical properties of natural stimuli.

We present a real-time model of learning in the auditory cortex that is trained using real-world stimuli. The system consists of a peripheral and a central cortical network of spiking neurons. The synapses formed by peripheral neurons on the central ones are subject to synaptic plasticity. We implemented a biophysically realistic learning rule that depends on the precise temporal relation of pre- and postsynaptic action potentials. We demonstrate that this biologically realistic real-time neuronal system forms stable receptive fields that accurately reflect the spectral content of the input signals and that the size of these representations can be biased by global signals acting on the local learning mechanism. In addition, we show that this learning mechanism shows fast acquisition and is robust in the presence of large imbalances in the probability of occurrence of individual stimuli and noise.

Neurons in mammalian cerebral cortex combine specific responses with respect to some stimulus features with invariant responses to other stimulus features. For example, in primary visual cortex, complex cells code for orientation of a contour but ignore its position to a certain degree. In higher areas, such as the inferotemporal cortex, translation-invariant, rotation-invariant, and even view point-invariant responses can be observed. Such properties are of obvious interest to artificial systems performing tasks like pattern recognition. It remains to be resolved how such response properties develop in biological systems. Here we present an unsupervised learning rule that addresses this problem. It is based on a neuron model with two sites of synaptic integration, allowing qualitatively different effects of input to basal and apical dendritic trees, respectively. Without supervision, the system learns to extract invariance properties using temporal or spatial continuity of stimuli. Furthermore, top-down information can be smoothly integrated in the same framework. Thus, this model lends a physiological implementation to approaches of unsupervised learning of invariant-response properties.

Electrophysiological research on the properties of the apical dendrites of cortical deep layer pyramidal cells suggests that it acts, in addition to the soma, as a second site of synaptic integration. Each site integrates input from a subset of synapses and is able to generate regenerative potentials. The sites exchange information in stereotyped ways: Signals from the soma are transmitted to the apical dendrite via actively back-propagating dendritic action potentials. Slow regenerative calcium spikes transmit information from the apical dendrite to the soma. These calcium spikes lead to a strong and prolonged depolarization of the cell generating a burst of action potentials. This paper analyzes how the system learns if these calcium spikes trigger hebbian learning at active synapses. A cell is now described by two main variables the mean activity and the mean potential at the apical dendrite with the first variable defining the input to cells downstream and the latter what the cell learns. A system results where neurons learn to respond to those features that are correlated with activity on the higher layer, this property is similar to maximizing the mutual information between input and higher areas. Furthermore it learns invariances exploiting a spatial smoothness criterion.

Mechanisms influencing learning in neural networks are usually investigated on either a local or a global scale. The former relates to synaptic processes, the latter to unspecific modulatory systems. Here we study the interaction of a local learning rule that evaluates coincidences of pre- and postsynaptic action potentials and a global modulatory mechanism, such as the action of the basal forebrain onto cortical neurons. The simulations demonstrate that the interaction of these mechanisms leads to a learning rule supporting fast learning rates, stability, and flexibility. Furthermore, the simulations generate two experimentally testable predictions on the dependence of backpropagating action potential on basal forebrain activity and the relative timing of the activity of inhibitory and excitatory neurons in the neocortex.

Neuroscience is progressing vigorously, and knowledge at different levels of description is rapidly accumulating. To establish relationships between results found at these different levels is one of the central challenges. In this simulation study, we demonstrate how microscopic cellular properties, taking the example of the action of modulatory substances onto the membrane leakage current, can provide the basis for the perceptual functions reflected in the macroscopic behavior of a cortical network. In the first part, the action of the modulatory system on cortical dynamics is investigated. First, it is demonstrated that the inclusion of these biophysical properties in a model of the primary visual cortex leads to the dynamic formation of synchronously active neuronal assemblies reflecting a context-dependent binding and segmentation of image components. Second, it is shown that the differential regulation of the leakage current can be used to bias the interactions of multiple cortical modules. This allows the flexible use of different feature domains for scene segmentation. Third, we demonstrate how, within the proposed architecture, the mapping of a moving stimulus onto the spatial dimension of the network results in an increased speed of synchronization. In the second part, we demonstrate how the differential regulation of neuromodulatory activity can be achieved in a self-consistent system. Three different mechanisms are described and investigated. This study thus demonstrates how a modulatory system, affecting the biophysical properties of single cells, can be used to achieve context-dependent processing at the system level.

Recent work suggests that synchronization of neuronal activity could serve to define functionally relevant relationships between spatially distributed cortical neurons. At present, it is not known to what extent this hypothesis is compatible with the widely supported notion of coarse coding, which assumes that features of a stimulus are represented by the graded responses of a population of optimally and suboptimally activated cells. To resolve this issue we investigated the temporal relationship between responses of optimally and suboptimally stimulated neurons in area 17 of cat visual cortex. We find that optimally and suboptimally activated cells can synchronize their responses with a precision of a few milliseconds. However, there are consistent and systematic deviations of the phase relations from zero phase lag. Systematic variation of the orientation of visual stimuli shows that optimally driven neurons tend to lead over suboptimally activated cells. The observed phase lag depends linearly on the stimulus orientation and is, in addition, proportional to the difference between the preferred orientations of the recorded cells. Similar effects occur when testing the influence of the movement direction and the spatial frequency of visual stimuli. These results suggest that binding by synchrony can be used to define assemblies of neurons representing a coarse-coded stimulus. Furthermore, they allow a quantitative test of neuronal network models designed to reproduce physiological results on stimulus-specific synchronization.

A temporal structure of neuronal activity has been suggested as a potential mechanism for defining cell assemblies in the brain. This concept has recently gained support by the observation of stimulus-dependent oscillatory activity in the visual cortex of the cat. Furthermore, experimental evidence has been found showing the formation and segregation of synchronously oscillating cell assemblies in response to various stimulus conditions. In previous work, we have demonstrated that a network of neuronal oscillators coupled by synchronizing and desynchronizing delay connections can exhibit a temporal structure of responses, which closely resembles experimental observations. In this paper, we investigate the self-organization of synchronizing and desynchronizing coupling connections by local learning rules. Based on recent experimental observations, we modify synchronizing connections according to a two-threshold learning rule, involving synaptic potentiation and depression. This rule is generalized to its functional inverse for weight changes of desynchronizing connections. We show that after training, the resulting network exhibits stimulus-dependent formation and segregation of oscillatory assemblies in agreement with the experimental data. These results indicate that local learning rules during ontogenesis can suffice to develop a connectivity pattern in support of the observed temporal structure of stimulus responses in cat visual cortex.

Current concepts in neurobiology of vision assume that local object features are represented by distributed neuronal populations in the brain. Such representations can lead to ambiguities if several distinct objects are simultaneously present in the visual field. Temporal characteristics of the neuronal activity have been proposed as a possible solution to this problem and have been found in various cortical areas. In this paper we introduce a delayed nonlinear oscillator to investigate temporal coding in neuronal networks. We show synchronization within two-dimensional layers consisting of oscillatory elements coupled by excitatory delay connections. The observed correlation length is large compared to coupling length. Following the experimental situation, we then demonstrate the response of such layers to two short stimulus bars of varying gap distance. Coherency of stimuli is reflected by the temporal correlation of the responses, which closely resembles the experimental observations.

Recent theoretical and experimental work suggests a temporal structure of neuronal spike activity as a potential mechanism for solving the binding problem in the brain. In particular, recordings from cat visual cortex demonstrate the possibility that stimulus coherency is coded by synchronization of oscillatory neuronal responses. Coding by synchronized oscillatory activity has to avoid bulk synchronization within entire cortical areas. Recent experimental evidence indicates that incoherent stimuli can activate coherently oscillating assemblies of cells that are not synchronized among one another. In this paper we show that appropriately designed excitatory delay connections can support the desynchronization of two-dimensional layers of delayed nonlinear oscillators. Closely following experimental observations, we then present two examples of stimulus-dependent assembly formation in oscillatory layers that employ both synchronizing and desynchronizing delay connections: First, we demonstrate the segregation of oscillatory responses to two overlapping but incoherently moving stimuli. Second, we show that the coherence of movement and location of two stimulus bar segments can be coded by the correlation of oscillatory activity.

Nonlinear units with delayed coupling are used as a representation of elementary neuronal oscillators. These basic elements are then coupled through delay connections to form an oscillatory network. Two classes of synchronizing and desynchronizing coupling connections within networks of oscillatory units are studied. With these types of connections, stimulus coherency leads to synchronization of the activated neuronal oscillators. A contiguous stimulus is shown to be coded by an assembly of oscillators active with zero phase lag, as required by experimental data. Furthermore, activities resulting from different stimuli lead to the formation of distinct oscillating assemblies. Two different, but overlapping, stimuli become represented by two different, coherently oscillating cell assemblies whose repetitive activities are no longer synchronized. This again agrees well with experimental observations