S8 – Neural Computations in the Brain: Bridging Theory and Experiments

Understanding the mechanisms employed by the brain to acquire, process and store information is a complex task which benefits from an interdisciplinary approach, combining empirical knowledge and expertise from the biological sciences with analytic and modeling tools commonly employed in the exact sciences. This includes tools from dynamical systems, signal processing, machine learning, and probabilistic modeling among others. This symposium brings together four scientists who apply a wide range of tools in order to further our understanding of perception, representations, belief formation, and behavioral outputs.

Note: A special emphasis has been put on diversity, not only in terms of gender but also trying to bring together well established scientists with young investigators.

Understanding the cortical representation of vocal communication signals is a major challenge for behavioral neuroscience. In this work, we report evidence of low dimensional neural activity in nucleus HVC (proper name, telencephalic sensorimotor area), during the song production in adult male canaries (Serinus canaria). We recorded multi-unit neural activity in multiple HVC sites and used machine learning techniques to process the data. We unveiled a low dimensional representation of the neural recordings analyzing the modes of the latent space of an auto-encoder. Overall, our results show that the rhythmic features of the vocal behaviors are represented in a telencephalic region of canaries.

Ana Amador
anita@df.uba.ar
Dynamical Systems Lab, DF, FCEN-UBA
https://www.df.uba.ar/es/component/researchers/miembro/68-Ana_Amador

The Bayesian theory of visual perception assumes that, given a stimulus, the brain performs probabilistic inference to estimate probabilistic distributions over unobservable variables. This process involves combining sensory information with previous expectations captured by a prior distribution. To understand how this process might occur in the cortex, we train artificial neural networks for a perceptual task: performing Bayesian inference in the context of natural images. In this case, we train Variational Autoencoders, which simultaneously learn a generative model of image patches alongside the corresponding inference model. We show that, under the requirement of optimal inference and using sparse activations, representations similar to those observed in the visual cortex emerge within the network. Notably, when an explicit contrast variable is included in the model, the network is able to not only correctly represent mean estimates about these over unobservable variables but also the level of remaining uncertainty after the observation.

Josefina Catoni
jcatoni@sinc.unl.edu.ar
sinc(i), CONICET-UNL
https://sinc.unl.edu.ar/staff/josefina-catoni/

Misinformation harms society by affecting citizens’ beliefs and behavior. Recent research has shown that partisanship and cognitive reflection (i.e. engaging in analytical thinking) play key roles in the acceptance of misinformation. However, the relative importance of these factors remains a topic of ongoing debate. In a registered report study, we tested four hypotheses on the relationship between each factor and the belief in statements made by Argentine politicians. Participants (N = 1353) classified fact-checked political statements as true or false, completed a cognitive reflection test, and reported their voting preferences. Using Signal Detection Theory and Bayesian modeling, we found a positive association between political concordance and overall belief, and additionally, that individuals with higher cognitive reflection exhibited greater skepticism, improved truth discernment, but also heightened partisan bias. Our results highlight the need to further investigate the relationship between cognitive reflection and partisanship in different contexts and formats.

Guillermo Solovey
gsolovey@gmail.com
Instituto de Cálculo, CONICET-UBA
https://www.ic.fcen.uba.ar/institucional/integrantes/solovey

Traditionally, semantic processes have been associated with the activation of multimodal temporal cortices. In recent years, substantial research has expanded this notion by highlighting the significant involvement of modality-specific systems. Yet, what the scope of such embodied phenomena is across conceptual categories remains to be explored. Here, I will introduce a framework combining lesion models with EEG and iEEG to capture key correlates of action, emotional, and negation concepts. Strategically, we focus on persons with movement disorders (Parkinson’s disease), socio-behavioral disruptions (behavioral variant frontotemporal dementia), and depth electrodes due to refractory epilepsy. Converging evidence from ERP and oscillatory measures show that these action, emotional, and negation concepts distinctly modulate canonical motor, affective, and inhibitory electrophysiological mechanisms, respectively. Signatures of normal and abnormal processing of these categories span early (< 300 ms) and late (> 300 ms) windows, attesting to the temporal ubiquity of embodied reactivations in the human brain. Altogether, these results illuminate the role of modality specific systems in semantic processing, while pointing to new electrophysiological indexes for disease characterization and monitoring.

Mariano Nicolás Díaz Rivera
mdiazrivera@udesa.edu.ar
CNC-UdeSA
https://udesa.edu.ar/cuerpo-docente/mariano-diaz-rivera