Seminars

It has long been appreciated (and celebrated) that certain species have sensory capabilities that humans do not share, for example polarization, ultraviolet, and infrared vision. What is less appreciated however, is that our position as terrestrial human scientists can significantly affect our study of animal senses and signals, even within modalities that we do share. For example, our acute vision can lead us to over-interpret the relevance of fine patterns in animals with coarser vision, and our Cartesian heritage as scientists can lead us to divide sensory modalities into orthogonal parameters (e.g. hue and brightness for color vision), even though this division may not exist within the animal itself. This talk examines two cases from marine visual ecology where a reconsideration of our biases as sharp-eyed Cartesian land mammals can help address questions in visual ecology. The first case examines the enormous variation in visual acuity among animals with image-forming eyes, and focuses on how acknowledging the typically poorer resolving power of animals can help us interpret the function of color patterns in cleaner shrimp and their client fish. The second case examines the how the typical human division of polarized light stimuli into angle and degree of polarization is problematic, and how a physiologically relevant interpretation is both closer to the truth and resolves a number of issues, particularly when considering the propagation of polarized light

I will present field experiments on vervet monkeys testing physical and social cognition, with a focus on social learning. The understanding of the emergence of cultural behaviours in animals has advanced significantly with contributions from complementary approaches: natural observations and controlled field experiments. Experiments with wild vervet monkeys highlight that monkeys are selective about ‘who’ they learn from socially and that they will abandon personal foraging preferences in favour of group norms new to them. The reported findings highlight the feasibility to study cognition under field conditions.

Large genomic studies of individuals with autism spectrum disorders (ASD) have revealed approximately 100-200 high risk genes. However, whether these genes function in similar or different signaling networks in brain cells (neurons) remains poorly studied. We are using proteomic technology to build an ASD-associated signaling network map as a resource for the Autism research community. This resource can be used to study Autism risk genes and understand how pathways are convergent, and how patient mutations change the interaction profile. In this presentation, we will present how we developed a pipeline using neurons to build protein-protein interaction profiles. We detected previously unknown interactions between different ASD risk genes that have never been linked together before, and for some genes, we identified new signaling pathways that have not been previously reported. This resource will be available to the research community and will foster collaborations between ASD researchers to help accelerate therapeutics for ASD and related disorders.

In recent years, artificial neural networks have demonstrated human-like or super-human performance in many tasks including image or speech recognition, natural language processing (NLP), playing Go, chess, poker and video-games. One remarkable feature of the resulting models is that they can develop very intuitive latent representations of their inputs. In these latent spaces, simple linear operations tend to give meaningful results, as in the well-known analogy QUEEN-WOMAN+MAN=KING. We postulate that human brain representations share essential properties with these deep learning latent spaces. To verify this, we test whether artificial latent spaces can serve as a good model for decoding brain activity. We report improvements over state-of-the-art performance for reconstructing seen and imagined face images from fMRI brain activation patterns, using the latent space of a GAN (Generative Adversarial Network) model coupled with a Variational AutoEncoder (VAE). With another GAN model (BigBiGAN), we can decode and reconstruct natural scenes of any category from the corresponding brain activity. Our results suggest that deep learning can produce high-level representations approaching those found in the human brain. Finally, I will discuss whether these deep learning latent spaces could be relevant to the study of consciousness.

Neurons and neural circuits can produce stereotyped and reliable output activity on the basis of highly variable cellular, synaptic, and circuit properties. This is crucial for proper nervous system function throughout an animal’s life in the face of growth, perturbations, and molecular turnover. But how can reliable output arise from neurons and synapses whose parameter vary between individuals in a population, and within an individual over time? I will review how a combination of experimental and computational methods can be used to examine how neuron and network function depends on the underlying parameters, such as neuronal membrane conductances and synaptic strengths. Within the high-dimensional parameter space of a neural system, the subset of parameter combinations that produce biologically functional neuron or circuit activity is captured by the notion of a ‘solution space’. I will describe solution space structures determined from electrophysiology data, ion channel expression levels across populations of neurons and animals, and computational parameter space explorations. A key finding centers on experimental and computational evidence for parameter correlations that give structure to solution spaces. Computational modeling suggests that such parameter correlations can be beneficial for constraining neuron and circuit properties to functional regimes, while experimental results indicate that neural circuits may have evolved to implement some of these beneficial parameter correlations at the cellular level. Finally, I will review modeling work and experiments that seek to illuminate how neural systems can homeostatically navigate their parameter spaces to stably remain within their solution space and reliably produce functional output, or to return to their solution space after perturbations that temporarily disrupt proper neuron or network function.

In the Learning Salon, we will discuss the similarities and differences between biological and machine learning, including individuals with diverse perspectives and backgrounds, so we can all learn from one another.

Rajanlab designs neural network models constrained by experimental data, and reverse engineers them to figure out how brain circuits function in health and disease. Recently, we have been developing a powerful new theory-based framework for “in-vivo tract tracing” from multi-regional neural activity collected experimentally. We call this framework CURrent-Based Decomposition (CURBD). CURBD employs recurrent neural networks (RNNs) directly constrained, from the outset, by time series measurements acquired experimentally, such as Ca2+ imaging or electrophysiological data. Once trained, these data-constrained RNNs let us infer matrices quantifying the interactions between all pairs of modeled units. Such model-derived “directed interaction matrices” can then be used to separately compute excitatory and inhibitory input currents that drive a given neuron from all other neurons. Therefore different current sources can be de-mixed – either within the same region or from other regions, potentially brain-wide – which collectively give rise to the population dynamics observed experimentally. Source de-mixed currents obtained through CURBD allow an unprecedented view into multi-region mechanisms inaccessible from measurements alone. We have applied this method successfully to several types of neural data from our experimental collaborators, e.g., zebrafish (Deisseroth lab, Stanford), mice (Harvey lab, Harvard), monkeys (Rudebeck lab, Sinai), and humans (Rutishauser lab, Cedars Sinai), where we have discovered both directed interactions brain wide and inter-area currents during different types of behaviors. With this powerful framework based on data-constrained multi-region RNNs and CURrent Based Decomposition (CURBD), we ask if there are conserved multi-region mechanisms across different species, as well as identify key divergences.

Initial social perceptions are often thought to reflect direct “read outs” of facial features. Instead, we outline a perspective whereby initial perceptions emerge from an automatic yet gradual process of negotiation between the perceptual cues inherent to a person (e.g., facial cues) and top-down social cognitive processes harbored within perceivers. This perspective argues that perceivers’ social-conceptual knowledge in particular can have a fundamental structuring role in perceptions, and thus how we think about social groups, emotions, or personality traits helps determine how we visually perceive them in other people. Integrative evidence from real-time behavioral paradigms (e.g., mouse-tracking), multivariate fMRI, and computational modeling will be discussed. Together, this work shows that the way we use facial cues to categorize other people into social groups (e.g., gender, race), perceive their emotion (e.g., anger), or infer their personality (e.g., trustworthiness) are all fundamentally shaped by prior social-conceptual knowledge and stereotypical assumptions. We find that these top-down impacts on initial perceptions are driven by the interplay of higher-order prefrontal regions involved in top-down predictions and lower-level fusiform regions involved in face processing. We argue that the perception of social categories, emotions, and traits from faces can all be conceived as resulting from an integrated system relying on domain-general cognitive properties. In this system, both visual and social cognitive processes are in a close exchange, and initial social perceptions emerge in part out of the structure of social-conceptual knowledge.

1. Embedding is all you need: A machine learning way to analyse white matter tractography streamlines - Dr Shenjun Zhong, Monash Biomedical Imaging Embedding white matter streamlines with various lengths into fixed-length latent vectors enables users to analyse them with general data mining techniques. However, finding a good embedding schema is still a challenging task as the existing methods based on spatial coordinates rely on manually engineered features, and/or labelled dataset. In this webinar, Dr Shenjun Zhong will discuss his novel deep learning model that identifies latent space and solves the problem of streamline clustering without needing labelled data. Dr Zhong is a Research Fellow and Informatics Officer at Monash Biomedical Imaging. His research interests are sequence modelling, reinforcement learning and federated learning in the general medical imaging domain. 2. Application of artificial intelligence in correcting motion artifacts and reducing scan time in MRI - Dr Kamlesh Pawar, Monash Biomedical imaging Magnetic Resonance Imaging (MRI) is a widely used imaging modality in clinics and research. Although MRI is useful it comes with an overhead of longer scan time compared to other medical imaging modalities. The longer scan times also make patients uncomfortable and even subtle movements during the scan may result in severe motion artifact in the images. In this seminar, Dr Kamlesh Pawar will discuss how artificial intelligence techniques can reduce scan time and correct motion artifacts. Dr Pawar is a Research Fellow at Monash Biomedical Imaging. His research interest includes deep learning, MR physics, MR image reconstruction and computer vision.

The ‘Hunt the Sex Difference’ agenda has informed brain research for decades, if not centuries. This talk aims to demonstrate how a fixed belief in differences between ‘male’ and ‘female’ brains can narrow and even distort the research process. This can include the questions that are asked, the methodology selected and the analytical pipeline. It can also powerfully inform the interpretation of results and the ‘spin’ used in the public communication of such research.

Progress in psychology has been frustrated by challenges concerning replicability, generalizability, strategy selection, inferential reproducibility, and computational reproducibility. Although often discussed separately, I argue that these five challenges share a common cause: insufficient investment of resources into the typical psychology study. I further suggest that big team science can help address these challenges by allowing researchers to pool their resources to efficiently and drastically increase the amount of resources available for a single study. However, the current incentives, infrastructure, and institutions in academic science have all developed under the assumption that science is conducted by solo Principal Investigators and their dependent trainees. These barriers must be overcome if big team science is to be sustainable. Big team science likely also carries unique risks, such as the potential for big team science institutions to monopolize power, become overly conservative, make mistakes at a grand scale, or fail entirely due to mismanagement and a lack of financial sustainability. I illustrate the promise, barriers, and risks of big team science with the experiences of the Psychological Science Accelerator, a global research network of over 1400 members from 70+ countries.

In this talk series, addiction neuroscientists from across the world share their personal stories/experiences on the beauty of addiction neuroscience and how/why they have decided to invest their scientific life in this field. We hope that this talk series would encourage and support a new generation of young and passionate addiction neuroscientists in different countries to revolutionize the field of addiction medicine.