Integrated Calendar

The dynamics of learning in natural and artificial environments is a problem of great interest to both neuroscientists and artificial intelligence experts. However, standard analyses of animal training data either treat behavior as fixed, or track only coarse performance statistics (e.g., accuracy and bias), providing limited insight into the dynamic evolution of behavioral strategies over the course of learning. To overcome these limitations, we propose a dynamic psychophysical model that efficiently tracks trial-to-trial changes in behavior over the course of training. In this talk, I will describe recent work based on a dynamic logistic regression model that captures the time-varying dependencies of behavior on stimuli and other task covariates. We applied our method to psychophysical data from both human subjects and rats learning a sensory discrimination task. We successfully tracked the dynamics of psychophysical weights during training, capturing day-to-day and trial-to-trial fluctuations in behavioral strategy. We leverage the model's flexibility model to investigate why rats frequently make mistakes on easy trials, demonstrating that so-called "lapses" often arise from sub-optimal weighting of task covariates. Finally, I will describe recent work on adaptive optimal training, which combines ideas from reinforcement learning and adaptive experimental design to formulate methods for inferring animal learning rules from behavior, and using these rules to speed up animal training.

Biomolecular imaging at the preclinical stage is an essential tool in various biomedical research areas such as immunology, oncology or neurology. Among all modalities available to date, optical imaging techniques play a central role, while fluorescence, in particular in the NIR region of the spectrum, provides high sensitivity and high specificity with relatively cheap instrumentation. Several whole-body optical pre-clinical NIR imaging systems are commercially available. Instruments using continuous wave (CW or time-independent) illumination allow basic small animal imaging at low cost. However, CW techniques cannot provide fluorescence lifetime contrast, which allows to probe the microenvironment and affords an increased multiplexing power. In the first part of my talk I will introduce our single photon, time-gated, phasor-based fluorescence lifetime Imaging method which circumvents limitations of conventional techniques in speed, specificity and ease of use, using fluorescent lifetime as the main contrast mechanism.
In the second part of my talk I will present the tracking and multiplexing of two different cell populations, based on their different lifetimes (following their fluorescent dyes-loading). Despite major advantages of optical based NIR imaging, the reason that NIR imagers are not clinically used, is that only very few such fluorescent molecules absorb and emit in the NIR (or in the shortwave infrared, SWIR region), and even fewer have favorable biological properties (and FDA approval). I will introduce small lung cancer and dendritic cells tracking using small polyethylene glycol/phosphatidylethanolamine (PEG–PE) micelles loaded with NIR dyes (using commercial dyes as well as dyes synthesized in Prof. Sletten’s lab, UCLA Chemistry Dept.). Micelles’ endocytosis into cells affords efficient loading and exhibits strong bio stability, enabling to track the loaded cells for several days using these formulations, even though dyes were diluted by cells division (leading to reduced dye concentration within the dividing cells). Moreover, fluorescent lifetime contrast (achieved through our time-gated imaging method), significantly improved these cells detection.
These advances in NIR fluorescence based imaging open up new avenues toward NIR and SWIR imaging for biomedical applications, such as tracking and monitoring cells during immunotherapy and/or drug delivery (treatment monitoring) for various types of disease.

An important problem in neuroscience is to identify low-dimensional structure underlying noisy, high-dimensional spike trains. In this talk, I will discuss recent advances for tackling this problem in single and multi-region neural datasets. First, I will discuss the Gaussian Process Latent Variable Model with Poisson observations (Poisson-GPLVM), which seeks to identify a low-dimensional nonlinear manifold from spike train data. This model can successfully handle datasets that appear high-dimensional with linear dimensionality reduction methods like PCA, and we show that it can identify a 2D spatial map underlying hippocampal place cell responses from their spike trains alone. Second, I will discuss recent extensions to Poisson-spiking Gaussian Process Factor Analysis (Poisson-GPFA), which incorporates separate signal and noise dimensions as well as a multi-region model with coupling between latent variables governing activity in different regions. This model provides a powerful tool for characterizing the flow of signals between brain areas, and we illustrate its applicability using multi-region recordings from mouse visual cortex.

Processing of a visual stimulus depends on previous and surrounding stimulations. For example, how an orientation detail is perceived depends on previous and surrounding orientation content. The influence of such context, temporal and spatial, is postulated to be beneficial, but the involved mechanism(s) as well as the behavioral relevance are not fully understood. Here, using behavioral experiments that measure how context integrates in space and time, we argue that context changes how statistical decisions are made by the visual system. Most importantly, we find that several context-dependent perceptual biases, such as visual illusions and aftereffects, are much reduced with increasing reaction time. To account for this, we consider a simple yet general explanation: prior and noisy decision-related evidence are integrated serially, with evidence and noise accumulating over time (as in the standard drift diffusion model). With time, owing to noise accumulation, the prior effect is predicted to diminish. This theory suggests a single-process alternative to the intuitive notion of dual brain systems (the so-called System 1 and System 2), and quantitatively predicts several known properties of perceptual bias, such as the order-of-magnitude variation in measured bias magnitudes between individuals.