Controlled variations in stimulus similarity during learning determine visual discrimination capacity in freely moving mice

Visuomotor predictors of interception

Intercepting moving targets is a fundamental skill in human behavior, influencing various domains such as sports, gaming, and other activities. In these contexts, precise visual processing and motor control are crucial for adapting and navigating effectively. Nevertheless, there are still some gaps in our understanding of how these elements interact while intercepting a moving target. This study explored the dynamic interplay among eye movements, pupil size, and interceptive hand movements, with visual and motion uncertainty factors. We developed a simple visuomotor task in which participants used a joystick to interact with a computer-controlled dot that moved along two-dimensional trajectories. This virtual system provided the flexibility to manipulate the target's speed and directional uncertainty during chase trials. We then conducted a geometric analysis based on optimal angles for each behavior, enabling us to distinguish between simple tracking and predictive trajectories that anticipate future positions of the moving target. Our results revealed the adoption of a strong interception strategy as participants approached the target. Notably, the onset and amount of optimal interception strategy depended on task parameters, such as the target's speed and frequency of directional changes. Furthermore, eye-tracking data showed that participants continually adjusted their gaze speed and position, continuously adapting to the target's movements. Finally, in successful trials, pupillary responses predicted the amount of optimal interception strategy while exhibiting an inverse relationship in trials without collisions. These findings reveal key interactions among visuomotor parameters that are crucial for solving complex interception tasks.

Encoding luminance surfaces in the visual cortex of mice and monkeys: difference in responses to edge and center

Luminance and spatial contrast provide information on the surfaces and edges of objects. We investigated neural responses to black and white surfaces in the primary visual cortex (V1) of mice and monkeys. Unlike primates that use their fovea to inspect objects with high acuity, mice lack a fovea and have low visual acuity. It thus remains unclear whether monkeys and mice share similar neural mechanisms to process surfaces. The animals were presented with white or black surfaces and the population responses were measured at high spatial and temporal resolution using voltage-sensitive dye imaging. In mice, the population response to the surface was not edge-dominated with a tendency to center-dominance, whereas in monkeys the response was edge-dominated with a "hole" in the center of the surface. The population response to the surfaces in both species exhibited suppression relative to a grating stimulus. These results reveal the differences in spatial patterns to luminance surfaces in the V1 of mice and monkeys and provide evidence for a shared suppression process relative to grating.

Intersecting of circular apertures to measure integer and fractional topological charge of vortex beams

Diffraction patterns of optical vortex beams (VBs) by differently shaped apertures are used to determine their topological charge (TC). In this paper, we show by simulations and experiments that diffraction of a Laguerre-Gaussian (LG) beam by intersecting circular apertures can be used to reveal the TC. The presented aperture structure has the advantage of the measurement of fractional TC in addition to the integer, sensitivity to the sign of TC, and low sensitivity to adjusting apertures. Accordingly, in addition to the integer TC up to 8, the fractional TC is measured with a step of 0.1 by two intersecting circular apertures (TICA). By examining a wide range of similarity criteria between the diffraction pattern of the fractional TC and the pattern of the lower integer TC, three metrics for measuring the fractional TC are found. Furthermore, the determination of integer TC up to 6 for three intersecting circular apertures (THICA) is demonstrated.

Response Time Distributions and the Accumulation of Visual Evidence in Freely Moving Mice

Response time (RT) distributions are histograms of the observed RTs for discriminative choices, comprising a rich source of empirical information to study perceptual processes. The drift-diffusion model (DDM), a mathematical formulation predicting decision tasks, reproduces the RT distributions, contributing to our understanding of these processes from a theoretical perspective. Notably, although the mouse is a popular model system for studying brain function and behavior, little is known about mouse perceptual RT distributions, and their description from an information-accumulation perspective. We combined an automated visual discrimination task with pharmacological micro-infusions of targeted brain regions to acquire thousands of responses from freely-moving adult mice. Both choices and escape latencies showed a strong dependency on stimulus discriminability. By applying a DDM fit to our experimental data, we found that the rate of incoming evidence (drift rate) increased with stimulus contrast but was reversibly impaired when inactivating the primary visual cortex (V1). Other brain regions involved in the decision-making process, the posterior parietal cortex (PPC) and the frontal orienting fields (FOF), also influenced relevant parameters from the DDM. The large number of empirical observations that we collected for this study allowed us to achieve accurate convergence for the model fit. Therefore, changes in the experimental conditions were mirrored by changes in model parameters, suggesting the participation of relevant brain areas in the decision-making process. This approach could help interpret future studies involving attention, discrimination, and learning in adult mice.

Isomorphic decisional biases across perceptual tasks

Humans adjust their behavioral strategies to maximize rewards. However, in the laboratory, human decisional biases exist and persist in two alternative tasks, even when this behavior leads to a loss in utilities. Such biases constitute the tendency to choose one action over others and emerge from a combination of external and internal factors that are specific for each individual. Here, we explored the idea that internally-mediated decisional biases should stably occur and, hence, be reflected across multiple behavioral tasks. Our experimental results confirm this notion and illustrate how participants exhibited similar choice biases across days and tasks. Moreover, we show how side-choice behavior in a two alternative choice task served to identify participants, suggesting that individual traits could underlie these choice biases. The tasks and analytic tools developed for this study should become instrumental in exploring the interaction between internal and external factors that contribute to decisional biases. They could also serve to detect psychopathologies that involve aberrant levels of choice variability.

Distributed processing of side-choice biases

Psychophysics describes how variations in stimulus strength lead to changes in perceptual performance. Yet, the contribution of non-sensory information processing to perceptual decision making is still not fully understood. For instance, in two-alternative forced-choice tasks, observers can exhibit tendencies to choose more one alternative over another, with no apparent goal or function. Such choice biases are highly prevalent in mice and, in free-choice tasks, they are insensitive to changes in stimulus discriminability. Thus, a reasonable proposal is that these side-choice biases could derive from functional asymmetries in sensory processing, decision making, or both. Here, we explored how different circuits participate in the production of choice biases in adult mice. We found that the magnitude of the changes in biased choice behavior depended on the inactivated region. Indeed, contralateral, but not ipsilateral, inactivations of the primary visual and posterior parietal cortices reduced the probability of mice choosing their preferred side. In contrast, ipsilateral inactivations of the subtantia nigra pars reticulata and of the frontal orienting fields, reduced and increased the probabilities of mice choosing their preferred side, respectively. These results demonstrate that internal circuit processing contributes to side-choice behavior and illustrates how distinct brain regions could participate in producing normal to aberrant levels of choice variability.

A novel paradigm for assessing olfactory working memory capacity in mice

A decline in working memory (WM) capacity is suggested to be one of the earliest symptoms observed in Alzheimer's disease (AD). Although WM capacity is widely studied in healthy subjects and neuropsychiatric patients, few tasks are developed to measure this variation in rodents. The present study describes a novel olfactory working memory capacity (OWMC) task, which assesses the ability of mice to remember multiple odours. The task was divided into five phases: context adaptation, digging training, rule-learning for non-matching to a single-sample odour (NMSS), rule-learning for non-matching to multiple sample odours (NMMS) and capacity testing. During the capacity-testing phase, the WM capacity (number of odours that the mice could remember) remained stable (average capacity ranged from 6.11 to 7.00) across different testing sessions in C57 mice. As the memory load increased, the average errors of each capacity level increased and the percent correct gradually declined to chance level, which suggested a limited OWMC in C57 mice. Then, we assessed the OWMC of 5 × FAD transgenic mice, an animal model of AD. We found that the performance displayed no significant differences between young adult (3-month-old) 5 × FAD mice and wild-type (WT) mice during the NMSS phase and NMMS phase; however, during the capacity test with increasing load, we found that the OWMC of young adult 5 × FAD mice was significantly decreased compared with WT mice, and the average error was significantly increased while the percent correct was significantly reduced, which indicated an impairment of WM capacity at the early stage of AD in the 5 × FAD mice model. Finally, we found that FOS protein levels in the medial prefrontal cortex and entorhinal cortex after the capacity test were significantly lower in 5 × FAD than WT mice. In conclusion, we developed a novel paradigm to assess the capacity of olfactory WM in mice, and we found that OWMC was impaired in the early stage of AD.

Non-stationary Salience Processing During Perceptual Training in Humans

Performance in sensory tasks improves with practice. Some theories suggest that the generalization of learning depends on task difficulty. In consequence, most studies have focused on measuring learning specificity, and perceptual impact after training completes. However, how exactly sustained changes in task difficulty influence the learning curves and how this affects the efficiency of perceptual discrimination is not well understood. Here, we adapted a visual task for humans by creating monocular training programs with increasing (SIM_inc) and decreasing (SIM_dec) stimulus similarities. We found a marked improvement in all participants after 10 days of training, with an almost complete transfer of learning to the untrained eyes. Interestingly, the training paradigms led to drastically different learning curves for the SIM_inc and SIM_dec groups. The learning curves were best predicted by an associative learning model that allowed stimuli to gain or lose salience depending on how the subject's learned about them. On addition, a non-stationary sequential sampling model that jointly accounts for choice and RT distributions revealed that the SIM_inc group led to faster evidence accumulation rate relative to the SIM_dec group. Altogether, our results illustrate how different learning trajectories influenced attentional salience processing leading to distinctive stimulus processing efficiencies. This crucial interdependence determines how observers learn to guide their attention towards visual stimuli in search for a decision.

Adaptive Choice Biases in Mice and Humans

The contribution of non-sensory information processing to perceptual decision making is not fully understood. Choice biases have been described for mice and humans and are highly prevalent even if they decrease rewarding outcomes. Choice biases are usually reduced by discriminability because stimulus strength directly enables the adjustments in the decision strategies used by decision-makers. However, choice biases could also derive from functional asymmetries in sensory processing, decision making, or both. Here, we tested how particular experimental contingencies influenced the production of choice biases in mice and humans. Our main goal was to establish the tasks and methods to jointly characterize psychometric performance and innate side-choice behavior in mice and humans. We implemented forced and un-forced visual tasks and found that both species displayed stable levels of side-choice biases, forming continuous distributions from low to high levels of choice stereotypy. Interestingly, stimulus discriminability reduced the side-choice biases in forced-choice, but not in free-choice tasks. Choice biases were stable in appearance and intensity across experimental days and could be employed to identify mice and human participants. Additionally, side- and alternating choices could be reinforced for both mice and humans, implying that choice biases were adaptable to non-visual manipulations. Our results highlight the fact that internal and external elements can influence the production of choice biases. Adaptations of our tasks could become a helpful diagnostic tool to detect aberrant levels of choice variability.

Abolishing cAMP sensitivity in HCN2 pacemaker channels induces generalized seizures

Hyperpolarization-activated cyclic nucleotide-gated (HCN) channels are dually gated channels that are operated by voltage and by neurotransmitters via the cAMP system. cAMP-dependent HCN regulation has been proposed to play a key role in regulating circuit behavior in the thalamus. By analyzing a knockin mouse model (HCN2EA), in which binding of cAMP to HCN2 was abolished by 2 amino acid exchanges (R591E, T592A), we found that cAMP gating of HCN2 is essential for regulating the transition between the burst and tonic modes of firing in thalamic dorsal-lateral geniculate (dLGN) and ventrobasal (VB) nuclei. HCN2EA mice display impaired visual learning, generalized seizures of thalamic origin, and altered NREM sleep properties. VB-specific deletion of HCN2, but not of HCN4, also induced these generalized seizures of the absence type, corroborating a key role of HCN2 in this particular nucleus for controlling consciousness. Together, our data define distinct pathological phenotypes resulting from the loss of cAMP-mediated gating of a neuronal HCN channel.

Adrenergic Modulation of Visually-Guided Behavior

Iontophoretic application of norepinephrine (NE) into the primary visual cortex (V1) in vivo reduces spontaneous and evoked activity, without changing the functional selectivity of cortical units. One possible consequence of this phenomenon is that adrenergic receptors (ARs) regulate the signal-to-noise ratio (SNR) of neural responses in this circuit. However, despite such strong inhibitory action of NE on neuronal firing patterns in V1, its specific action on visual behavior has not been studied. Furthermore, the majority of observations regarding cortical NE from in vivo recordings have been performed in anesthetized animals and have not been tested behaviorally. Here, we describe how micro-infusion of AR agonists/antagonists into mouse V1 influences visually-guided behavior at different contrasts and spatial frequencies. We found that cortical activation of α₁- and β-AR produced a substantial reduction in visual discrimination performance at high contrasts and low spatial frequencies, consistent with a divisive effect. This reduction was reversible and was accompanied by a rise in escape latencies as well as an increase in the group averaged choice variance as a function of stimulus contrast. We conclude that pharmacological activation of cortical AR regulates visual perception and adaptive behavior through a divisive gain control of visual responses.

An Automated Water Task to Test Visual Discrimination Performance, Adaptive Strategies and Stereotyped Choices in Freely Moving Mice

We describe an automated training/testing system for adult mice that allows reliable quantification of visual discrimination capacities, adaptive swimming strategies, and stereotyped choices with minimal human intervention. The experimental apparatus consists of a hexagonal swimming pool with an internal decision zone leading to three interior arms with two software-controlled platforms inside of each arm. Each experimental trial consists in projecting a "positive" conditioned discriminative stimulus (SD) in one randomly chosen arm, whereas the other two arms project non-reinforced stimuli (the delta stimuli, SΔ). By employing a classical behavioral training schedule, the mice learn to swim toward the arm that displays the SD, because it predicts the presence of two elevated platforms located symmetrically to the left and right side of the projecting monitor. Separate behavioral components for discriminative and stereotyped choice behavior can be identified through this geometric arrangement. In addition, the projection in real-time of either static or dynamic visual stimuli allows the usage of training programs contingent on current behavioral performance. We validated the system by characterizing the visual acuity and contrast sensitivities in a group of trained mice. By employing pharmacological manipulations, we found that the mice required an intact functioning of the primary visual cortex (V1) to solve the hexagonal pool. Overall, the automated training system constitutes a reliable, rapid, and inexpensive method to quantify visual capacities of mice. It can be used to characterize visual and non-visual factors of choice behavior. It can also be combined with manipulations of visual experience and pharmacological micro-infusions to investigate integrated brain function and learning processes in adult mice over consecutive days.

Mice use robust and common strategies to discriminate natural scenes

Mice use vision to navigate and avoid predators in natural environments. However, their visual systems are compact compared to other mammals, and it is unclear how well mice can discriminate ethologically relevant scenes. Here, we examined natural scene discrimination in mice using an automated touch-screen system. We estimated the discrimination difficulty using the computational metric structural similarity (SSIM), and constructed psychometric curves. However, the performance of each mouse was better predicted by the mean performance of other mice than SSIM. This high inter-mouse agreement indicates that mice use common and robust strategies to discriminate natural scenes. We tested several other image metrics to find an alternative to SSIM for predicting discrimination performance. We found that a simple, primary visual cortex (V1)-inspired model predicted mouse performance with fidelity approaching the inter-mouse agreement. The model involved convolving the images with Gabor filters, and its performance varied with the orientation of the Gabor filter. This orientation dependence was driven by the stimuli, rather than an innate biological feature. Together, these results indicate that mice are adept at discriminating natural scenes, and their performance is well predicted by simple models of V1 processing.

Noise Improves Visual Motion Discrimination via a Stochastic Resonance-Like Phenomenon

The stochastic resonance (SR) is a phenomenon in which adding a moderate amount of noise can improve the signal-to-noise ratio and performance of non-linear systems. SR occurs in all sensory modalities including the visual system in which noise can enhance contrast detection sensitivity and the perception of ambiguous figures embedded in static scenes. Here, we explored how adding background white pixel-noise to a random dot motion (RDM) stimulus produced changes in visual motion discrimination in healthy human adults. We found that, although the average reaction times (RTs) remained constant, an intermediate level of noise improved the subjects’ ability to discriminate motion direction in the RDM task. The psychophysical responses followed an inverted U-like function of the input noise, whereas the incorrect responses with short RTs did not exhibit such modulation by external noise. Moreover, by applying stimulus and noisy signals to different eyes, we found that the SR phenomenon occurred presumably in the primary visual cortex, where these two signals first converge. Our results suggest that a SR-like phenomenon mediates the improvement of visual motion perception in the RDM task.

Associative Learning Through Acquired Salience

Most associative learning studies describe the salience of stimuli as a fixed learning-rate parameter. Presumptive saliency signals, however, have also been linked to motivational and attentional processes. An interesting possibility, therefore, is that discriminative stimuli could also acquire salience as they become powerful predictors of outcomes. To explore this idea, we first characterized and extracted the learning curves from mice trained with discriminative images offering varying degrees of structural similarity. Next, we fitted a linear model of associative learning coupled to a series of mathematical representations for stimulus salience. We found that the best prediction, from the set of tested models, was one in which the visual salience depended on stimulus similarity and a non-linear function of the associative strength. Therefore, these analytic results support the idea that the net salience of a stimulus depends both on the items' effective salience and the motivational state of the subject that learns about it. Moreover, this dual salience model can explain why learning about a stimulus not only depends on the effective salience during acquisition but also on the specific learning trajectory that was used to reach this state. Our mathematical description could be instrumental for understanding aberrant salience acquisition under stressful situations and in neuropsychiatric disorders like schizophrenia, obsessive-compulsive disorder, and addiction.

Optimization of visual training for full recovery from severe amblyopia in adults

The severe amblyopia induced by chronic monocular deprivation is highly resistant to reversal in adulthood. Here we use a rodent model to show that recovery from deprivation amblyopia can be achieved in adults by a two-step sequence, involving enhancement of synaptic plasticity in the visual cortex by dark exposure followed immediately by visual training. The perceptual learning induced by visual training contributes to the recovery of vision and can be optimized to drive full recovery of visual acuity in severely amblyopic adults.

Undesirable Choice Biases with Small Differences in the Spatial Structure of Chance Stimulus Sequences

In two-alternative discrimination tasks, experimenters usually randomize the location of the rewarded stimulus so that systematic behavior with respect to irrelevant stimuli can only produce chance performance on the learning curves. One way to achieve this is to use random numbers derived from a discrete binomial distribution to create a 'full random training schedule' (FRS). When using FRS, however, sporadic but long laterally-biased training sequences occur by chance and such 'input biases' are thought to promote the generation of laterally-biased choices (i.e., 'output biases'). As an alternative, a 'Gellerman-like training schedule' (GLS) can be used. It removes most input biases by prohibiting the reward from appearing on the same location for more than three consecutive trials. The sequence of past rewards obtained from choosing a particular discriminative stimulus influences the probability of choosing that same stimulus on subsequent trials. Assuming that the long-term average ratio of choices matches the long-term average ratio of reinforcers, we hypothesized that a reduced amount of input biases in GLS compared to FRS should lead to a reduced production of output biases. We compared the choice patterns produced by a 'Rational Decision Maker' (RDM) in response to computer-generated FRS and GLS training sequences. To create a virtual RDM, we implemented an algorithm that generated choices based on past rewards. Our simulations revealed that, although the GLS presented fewer input biases than the FRS, the virtual RDM produced more output biases with GLS than with FRS under a variety of test conditions. Our results reveal that the statistical and temporal properties of training sequences interacted with the RDM to influence the production of output biases. Thus, discrete changes in the training paradigms did not translate linearly into modifications in the pattern of choices generated by a RDM. Virtual RDMs could be further employed to guide the selection of proper training schedules for perceptual decision-making studies.

Stimulus similarity determines the prevalence of behavioral laterality in a visual discrimination task for mice

Animal choices depend on direct sensory information, but also on the dynamic changes in the magnitude of reward. In visual discrimination tasks, the emergence of lateral biases in the choice record from animals is often described as a behavioral artifact, because these are highly correlated with error rates affecting psychophysical measurements. Here, we hypothesized that biased choices could constitute a robust behavioral strategy to solve discrimination tasks of graded difficulty. We trained mice to swim in a two-alterative visual discrimination task with escape from water as the reward. Their prevalence of making lateral choices increased with stimulus similarity and was present in conditions of high discriminability. While lateralization occurred at the individual level, it was absent, on average, at the population level. Biased choice sequences obeyed the generalized matching law and increased task efficiency when stimulus similarity was high. A mathematical analysis revealed that strongly-biased mice used information from past rewards but not past choices to make their current choices. We also found that the amount of lateralized choices made during the first day of training predicted individual differences in the average learning behavior. This framework provides useful analysis tools to study individualized visual-learning trajectories in mice.