The list of the abstracts of participants' presentations. Each presentation should last up to 20' questions included.
The order of the presentations will be as here, starting at 9:30 on Tuesday the 3rd:
Solving the spatial navigation watermaze task: a challenge in computational neuroscience.
I will present two models from Foster, Dayan and Morris  in which place cells are used to perform a watermaze navigation task. Both models use temporal difference learning rules to estimate a value function (for a discounted reward) and a policy (best action to perform) in each state (place in the maze provided by place cells). The first model is able to successfully learn goal locations, though with only a gradual improvement in performance that does not capture the one-shot learning seen in real experiments when rats learn novel goal locations repeatedly [2,3]. The second model is an extension of the first to include learning of absolute spatial position and overcomes some of the limitations of the first model. I will present and discuss the methodologies and performance of both models and compare them to real laboratory data obtained from rats . I will conclude by discussing some of the future challenges of my phd, and in particular that of understanding one-shot learning.
All-optical manipulation of hippocampal place cells drives reward-associated behaviour during spatial navigation
Hippocampal place cells fire when an animal occupies a specific location and are thought to support navigation and spatial memory. However it is not yet known how place cells drive behaviour during navigation. To investigate this question, we used an "all-optical" approach combining two-photon in-vivo calcium imaging and two-photon holographic optogenetic stimulation in head-fixed mice performing a navigation task in a virtual reality environment. Mice had to walk through a virtual corridor, stop at a platform for 3 seconds and lick 3 times to trigger a reward delivery. Place cells covering the virtual space were functionally identified and grouped into reward-zone associated cells or non reward-zone associated cells. We then specifically activated either the reward-zone cells or non reward-zone cells each time the animal crossed a central location in the track to test whether this would retrieve the behaviour associated with the reward area. Our preliminary results suggests that the specific activation of reward-zone cells can drive reward-associated behaviour. This approach enables us to probe the causal relationship between hippocampal neural activity patterns and the retrieval of memory content to guide behaviour.
The ability to plan routes is an essential part of daily life and relies on the knowledge of the street network and the related environmental structure, such as boundaries and bottlenecks. Given the potential complexity, it is unlikely that humans compare all possible routes to find a suitable solution, as suggested by conventional models. Instead, the environmental structure can foster a hierarchical representation of the environment that can facilitate or impair the route planning process and lead to differences in planning times. In a study with licensed London taxi driver who have unique knowledge of the street network in London and can flexibly plan routes without consulting a visual map, we test such a hierarchical planning model (HPM). We hypothesize that route planning is related to the planning demands of the environment and can be predicted by the HPM. Audio‑recorded data was collected from N=20 taxi drivers who planned routes between 12 origin‑destination pairs. The initial planning times of the routes, as well as the individual recall times between streets were extracted. Preliminary analysis point towards systematic errors during route‑recall (e.g. forgotten street names) and a variation of planning and recall times in line with a hierarchical representation of the street network.
Integration of path integration and contextual inputs in the hippocampal memory network: approaching the problem through functional-network inference and continuous-attractor modeling
The hippocampus is known to store cognitive representations, or maps, that encode both positional and contextual information, critical for episodic memories and functional behavior. How path integration and contextual cues are dynamically combined and processed by the hippocampus to maintain these representations accurate over time remains unclear. To answer this question, we propose a two-way data analysis and modeling approach to CA3 multi-electrode recordings of a moving rat submitted to rapid changes of contextual (light) cues, triggering back-and-forth instabitilies between two cognitive representations (Jezek et al, Nature 478, p 246 (2011)). We develop a dual neural activity decoder, capable of independently identifying the recalled cognitive map at high temporal resolution (comparable to theta cycle) and the position of the rodent given a map. Remarkably, position can be reconstructed at any time with an accuracy comparable to fixed-context periods, even during highly unstable periods. These findings provide evidence for the capability of the hippocampal neural activity to maintain an accurate encoding of spatial and contextual variables, while one of these variables undergoes rapid changes independently of the other. To explain this result we introduce an attractor neural network model for the hippocampal activity that process inputs from external cues and the path integrator. Our model allows us to make predictions on the frequency of the cognitive map instability, its duration, and the detailed nature of the place-cell population activity, which are validated by a further analysis of the data. Our work therefore sheds light on the mechanisms by which the hippocampal network achieves and updates multi-dimensional neural representations from various input streams.
Central amygdala circuits regulate exploration in response to social signals about threat
To avoid danger, animals need to regulate exploration based on emotional signals emitted by conspecifics. In order to study this process, we used two experimental paradigms in which rats interact with a fear conditioned partner. In the first one (imminent danger) an 'observer' watched his cagemate (‘demonstrator’) undergoing contextual fear conditioning. In the second paradigm (remote danger) the observer could freely interact with a recently conditioned demonstrator in the safe environment of a home cage. Both paradigms elicit robust but different reactions of observer rats - freezing and rearing, respectively. They are also accompanied by different patterns of ultrasonic vocalizations (centered around 22 vs. 50 kHz). As central amygdala (CeA) is crucial for single subject defensive behaviors, we hypothesised that it also plays role in regulating reactions to social signals about threats. To verify this, we used viral vectors in which channelorhodopsin (ChR2) or halorhodosin (NpHR) sequence was linked to c-fos promoter. This approach allowed us to selectively manipulate the subpopulations of CeA cells which were activated by either type of social interaction. Twenty four hours after the behavioral paradigm, observer rats received light stimulation either in the open field or during a free interaction with the cagemate. The results indicate that the population of CeA cells activated by imminent danger paradigm controls passive defensive reactions – such as hiding in dark areas and reduced exploration - whereas the one activated by remote danger has an opposite role, promoting active reactions including rearing and escape from the light. Suprisingly, stimulating none of the populations had direct effect on the observer social behaviors, suggesting that the main function of the studied circuits is regulation of exploratory strategy.
Soledad Gonzalo Cogno
Theta phase resetting events during exploratory behavior
Among all frequency bands of Local Field Potential (LFP) that modulate the activity of neurons in the Entorhinal Cortex (EC), the theta band has by far the strongest influence. We investigated theta oscillations during several repetitions of stereotypical behavior in data from Kropff et al, 2015. We found that a phase resetting consistently occurred inside a window of 1 second after the beginning of each lap. During such events, the circular variance of the theta phase across trials decreased suddenly and significantly. The resetting events typically lasted around one second, beyond which cross-trial coherence was lost. This effect could potentially indicate the onset of a mechanism such as path integration. To further investigate this, we identified specific properties of the LFP that served as a footprint to detect the phase resetting on the basis of individual traces rather than averaging across trials, both in Flintstone car and open field experiments. In the Flintstone car experiment, phase resetting events occurred not only at the beginning of each lap, but also during slow running periods. They never took place during fast running periods. In open field experiments, they occurred throughout the environment, and were correlated with brief acceleration periods. We are currently analyzing the hypothesis that phase resetting events coincide with the resetting of the path integrator, which should erase the accumulated error of grid cells (Hardcastle et al, 2015). Pilot results suggest that this is the case. In sum, phase resetting could potentially represent a salient event during the processing of self location, and a key element to understand the interaction between different entorhinal cells that participate in spatial orientation.
Functional connectivity in the medial prefrontal cortex during wakefulness and sleep
Characterization of cell interactions is fundamental to the understanding of information processing in neural circuits. Among all possible brain states, wakefulness and sleep have been extensively studied. However, how neurons organize and operate in both states remains unclear. Here we investigated functional connectivity in a network of 32 neurons recorded in medial prefrontal cortex (Schwindel et al, 2014), and characterized the profiles of connectivity during wakefulness and sleep. To this end, we used graph theory and generalized linear models. We found that connections are stronger, more widespread and dense during sleep. We also found that the network architecture depends strongly on the brain state, being more efficient for the transmission of information during sleep, whereas the network organizes through hubs and clusters during wakefulness.
Kamil Filip Tomaszewski
Neural circuits and functional connectivity of fear memory extinction in αCaMKII autophosphorylation-deficient mice
Understanding how activity in neural circuits drives behavior is a fundamental problem in neuroscience. Here we presents a functional connectome of 24 brain regions underlying a specific behavior of contextual conditioned fear extinction in WT and αCaMKII autophosphorylation-deficient heterozygous mice (T286A +/-) after recent (1 day after training) and remote (30 days after training) memory. αCaMKII mutants were chosen, because it has been shown previously that αCaMKII autophosphorylation plays a pivotal role in synaptic plasticity as well, as in memory formation and extinction. To identify neuronal network of contextual conditioned fear extinction and how they are represented by network interactions between brain regions we applied immediate early gene c-Fos immunostaining to determine activity of brain regions and graph theoretical concepts to calculate inter-regional correlation (network). Based on the analysis of immediate early gene c-Fos patterns in neurons of the fear extinction circuit, we claim that extinction is expressed differently when autophosphorylation of αCaMKII is disrupted. Network analysis showed that T286A +/- mice have also different functional connectomes (i.e. nodes distribution, vertex density) and their network disassembly, creating smaller nets after remote memory extinction. Moreover interdependence of these two measures in mutant mice is correlating with their inability to acquire and/or retrieval of remote memory extinction, suggesting an important role of αCaMKII autophosphorylation in this phenomenon.
Metric of the color space mediated by adaptation process
In spatially uniform and temporally stationary conditions of luminosity, the ability of discrimination for trichromat humans depends on the compared colours. For example, it is pretty much easier to discriminate two orange chromas whose wavelengths differ in a quantity $delta lambda$, that two green chromas with the same difference. In previous work, we developed a theoretical model which explains this inhomogeneities in term of the physiological properties of the retinal cones. The absorption of photons is a stochastic process which variability depends on the wavelength. This variability puts a limit to the discrimination of the electrical signals that represent the two compared colours. There are recent experiments that extends the previous studies to the situation where illumination is not spatially uniform (the colour of the sorround is controlled in the experiment) , neither temporally stationary (the stimulus are shows for a brief period). The results show that the capacity of discriminating colour depends strongly on these conditions. On this work, we extend the previous theoretical models by including the spatial and temporal structure of the stimulis used in the more recent experiments. We show that, as response to the chroma of the surround, the visual system generates a representation such that the test colour moves away.
Involvement of retrosplenial cortex in spatial learning and navigation
The retrosplenial cortex (RSC) is a cortical area in the brain which is a part of a core network of brain regions involved in a range of cognitive functions including navigation in the environment, formation, storage and retrieval of spatial information. However its precise role in memory functioning has not been sufficiently explained. It has been postulated that RSC might be responsible for processing of visual information about the external environment in order to allow orientation on the "mental map" generated by hippocampus. The main goal of our studies is to check whether RSC neurons form a memory trace based on Fos-inducing plasticity. We use a rat animal model subjected to spatial memory training in a T-maze task, which requires the navigation based on external cues. Using controlled expression of viral opsins followed by the LED diodes implanted on the surface of retrosplenial cortex we want to activate or inhibit the entire neuronal population of this structure during specific phases of training. By comparing the learning curves of control and treated animals it will be possible to identify the stage of learning where the contribution of RSC is crucial for memory formation. We also use a Designer Receptors Exclusively Activated by Designer Drugs (DREADD) to detect whether the ligand-induced temporal inactivation of whole RSC influence solving the task. Using Fos protein level as an indirect marker of neuronal activity we would like to answer the question whether temporal inhibition of RSC affects different brain areas which are very likely to take a part in solving a behavioral task. Complementing the T maze behavioral tests of rats with optogenetic and/or chemogenetic manipulation of neuronal signal transduction can lead to better understanding of important elements of the memory system that support navigation and spatial memory.
The role of the amygdala in emotional contagion in a mouse model of autism spectrum disorder
The aim of study was to assess the emphatic abilities and the activity pattern within the amygdala of Fmr1KO(FVB) mice. To study empathic abilities we employed the social fear transfer paradigm, in which mice are housed in pairs for three weeks, one labelled an Observer, and the other a Demonstrator. In the test session, the Demonstrator is subjected to aversive stimuli outside of the home cage, while the Observer remains there undisturbed. Then, the Demonstrator returns to the home cage, where it can freely interact with the Observer. First ten minutes of interactions are recorded. Ninety minutes after the onset of the interaction animals were sacrificed for immunochemistry. The activity of the amygdala was assessed using double-staining immunohistochemistry, against c-Fos protein, a standard neuronal novelty marker, and against Corticotropin-releasing hormone (CRH), which is involved in stress response. Fmr1KO(FVB) mice show a certain degree of emotional contagion. They show interest in a stressed conspecific, but tend to self-groom more when exposed to such stimulus. The activity pattern of the amygdala in Observers from Control and Experimental group is similar, except for a few differences in distinct amygdalar nuclei. The majority of c-Fos positive neurons were also CRH positive. Fmr1KO(FVB) mice, an animal model for autism spectrum disorder, display moderate empathic responsivity combined with enhanced stereotypy upon exposure to a stressed cagemate. The observed amygdala activation reflects the changed behavior. It also confirms the role of CRH neuropeptide in emotional contagion involving aversive stimuli.
Neuronal correlates of the extinction memory retrieval
Extinction memory is susceptible to the time passage as well as to the change of the context. Recovery-from-extinction effects has been well described on the behavioural level; however the neuronal basis of this phenomenon has not yet been understood. Using different approaches (c-Fos mapping, functional anatomy tracing and optogenetics) we investigated the circuitry underlying successful and unsuccessful extinction memory retrieval. We have found that different kind of projections from the ventral hippocampus to prelimbic cortex underlie retrieval of fear and extinction memory.
Switching States of Neuronal Networks in the Brain
Oscillations are believed to play a role in flexible information routing between regions but their impact on the way in which information is processed locally within region is still unclear. Here we use an information theory approach to quantify the involvement of individual neurons in the primitive information processing operations of storage and sharing that can be viewed as generic building blocks of local circuit computations. We introduce the concept of a computing microstate, an epoch in which neurons are assigned to a consistent information processing role. Each microstate has associated state-specific computing hubs, neurons most strongly involved in sharing and storage operations. We find that there is a multiplicity of different computing microstates, and that discrete transitions between them cause neurons and computing hubs to switch their functional roles. Importantly, global oscillatory states (slow and theta oscillations during anesthesia or sleep) affect both the repertoire of available computing microstates and the complexity of their observed temporal sequences. This enables a democratic role sharing in which nearly half of the recorded neurons can act as a computing hub in at least some of the microstates. Together, this suggests that the roles played by neurons in local computations are not firmly hard-wired but rather emerge as the effect of rich collective dynamics, modulated by global oscillatory states.
Exploring dynamical states in a network of excitatory and inhibitory spiking neurons
There is a prominent difference between neural population activity from In vitro and In vivo recordings, one showing burst behavior, whereas the other shows ongoing firing. A major difference between these two networks is that In vitro systems do not receive external input as much as In vivo ones. In our research we studied whether these two states are realized in a well-established neural network model (Brunel, 2000), and under which conditions.
Optogenetic inhibition of somatostatin basal forebrain's cells increases cortical activity
The basal forebrain provides modulatory input to the cortex regulating brain states and cognitive processing. Somatostatin-expressing cells constitute a local GABAergic source known to functionally inhibit the major cortically-projecting cell types. However, it remains unclear if somatostatin cells can regulate the basal forebrain's synaptic output and thus control cortical dynamics. Here, I demonstrate in mice that somatostatin neurons regulate the corticopetal synaptic output of the basal forebrain impinging on cortical activity and behavior. Optogenetic inactivation of somatostatin neurons in vivo increased spiking of some basal forebrain cells, rapidly enhancing and desynchronizing neural activity in the prefrontal cortex, increasing gamma oscillations and locomotor activity in quiescent animals. Altogether, I provide physiological and behavioral evidence indicating that somatostatin cells are pivotal in gating the synaptic output of the basal forebrain, thus indirectly controlling cortical operations via both cholinergic and non-cholinergic mechanisms.
Shafa Jaffal and Hala Khalawi
Al Quds University
Palestinian Neuroscience Intiative
The Effects of Menstrual Hormonal Fluctuations onFeedback-Based Learning in Healthy Young Women
In this study, we use computer-based tasks to study cognitive function in relation to levels of hormonal fluctuation across the menstrual cycle of healthy undergraduate women at Al-Quds University, Palestine. We study basal ganglia and hippocampal dependent learning from generalization and positive/negative feedback as a function of estrogen and progesterone variations in the follicular, ovulatory, and luteal phases of the menstrual cycle. Furthermore, we will correlate cognitive and hormonal changes to levels of depression and anxiety across the menstrual cycle.