Top-down feedback

Predictive processing offers a unifying view of neural computation, proposing that brains continuously anticipate sensory input and update internal models based on prediction errors. In this talk, I will present converging evidence for the computational mechanisms underlying this framework across human neuroscience and deep neural networks. I will begin with recent work showing that large-scale distributed prediction-error encoding in the human brain directly predicts how sensory representations reorganize through predictive learning. I will then turn to PredNet, a popular predictive coding in

…and theoretical work suggests that top-down input to the apical dendrites of pyramidal cells may provide a solution to this deep credit assignment problem. Experiments have shown that dendritic-spike-dependent plasticity is promoted by top-down excitatory inputs carrying information about learning targets, and it is suppressed by lateral feedback inhibition. However, exist

Visual representations become progressively more abstract along the cortical hierarchy. These abstract representations define notions like objects and shapes but at the cost of spatial specificity. By contrast, low-level regions represent spatially local but simple input features. How do spatially non-specific representations of abstract concepts in high-level areas flexibly modulate the low-level sensory representations in appropriate ways to guide context-driven and goal-directed behaviors across various tasks? We present a biologically motivated and trainable neural network model of dyna

…In contrast, $modulatory$ top-down feedback [10] (i.e., altering spiking propensity without inducing spiking directly [11]) from frontal regions to EVC

Repetition suppression (RS), the diminished neural response to repeated stimuli, is a well-documented phenomenon potentially explained by predictive coding. This theory suggests that the brain anticipates incoming information, minimizing further processing for familiar stimuli. For example, fusiform face area (FFA) consistently shows decreased neural response to repeated presentation of the same faces. According to the traditional model of visual processing, information progresses from the lower brain regions' superficial cortical layers to the higher areas' middle layers. Within this framewor

…for synaptic eligibility traces and top-down feedback signals, among other factors. Here we study the potential contribution of another type of signaling that is being revealed in greater and greater fidelity by ongoing molecular and genomic

During behavior, sensation and action operate in closed-loop. Movements shape sensory input, and sensory inputs guide motor commands: where one looks determines what one sees. Through experience, the brain learns the reciprocal relationship between sensory inputs and movements to build internal models that accurately predict the sensory consequences of upcoming actions (sensorimotor predictions). This exchange of sensory inputs and egocentric expectations is at the core of active perception. Experimental investigation of this idea has been sparse and split between behavioral interrogation of s

The ability to flexibly respond to sensory cues in dynamic environments is essential to adaptive auditory-guided behaviors such as navigation and communication. How do neural circuits flexibly gate sensory information to select appropriate behavioral strategies based on sensory input and context? Auditory neural responses during behavior are diverse, ranging from highly-reliable ‘classical’ responses (i.e. robust, frequency-tuned cells) to irregular or seemingly random ‘non-classically responsive’ firing patterns (i.e., nominally non-responsive cells) that fail to demonstrate any significant t

A central problem in biological learning is how information about the outcome of a decision or behavior can be used to reliably guide learning across distributed neural circuits while obeying biological constraints. This “credit assignment” problem is commonly solved in artificial neural networks through supervised gradient descent and the backpropagation algorithm. In contrast, biological learning is typically modelled using unsupervised Hebbian learning rules. While these rules only use local information to update synaptic weights, and are sometimes combined with weight constraints to reflec

Perception is influenced by past choices. In combined discrimination-estimation experiments a categorical choice leads to two biases: (i) a choice-dependent confirmation bias in the continuous estimate of the stimulus average, and (ii) an overall decrease in the sensitivity to subsequent sensory evidence. It remains unknown what neural mechanisms give rise to these post-decision biases and whether continuous estimates and categorical choices are computed in a common cortical circuit. Here, we develop a neural network model that addresses these questions. We study the integration of continuous