Biology subjects

Furlong, P. M.

Publications and source records attributed to Furlong, P. M..

3 recordsLinked to original sources

bioRxiv2026

Neural Representational Geometry of Feature Binding Operations

The brain faces the feature binding problem: how are multiple stimulus features and variables combined into coherent representations that support flexible behavior? A key finding from neuroscience is that some brain regions employ factorized representations, where distinct features are encoded in neural state space in such a way that enables independent readout and robust generalization. Various algebraic operations have been proposed to model multi-variable representations, but despite extensive study of their theoretical properties (e.g., capacity, noise robustness), it remains unclear which operations produce the representational geometries observed in neural recordings. We systematically evaluate six binding operations implemented in recurrent spiking neural networks performing a working memory task. We find that only superposition and binding with slot-filler structure produce factorized geometry with favorable scaling, while the alternatives do not. These results provide a taxonomy linking algebraic binding operations to neural representational signatures, offering guidance for both computational modelers and experimentalists.

neuroscience↗

bioRxiv2026

Novel model combining intrinsic and learned behaviours captures divergent effects of dopaminergic drugs on different types of motivation

Motivation-based symptoms occur in an array of neurological and neurodegenerative disorders, including Schizophrenia and Parkinsons Disease. Unfortunately, there is, as yet, no agreed treatment approach. To establish a better understanding of motivation, and so highlight potential avenues for treatment, motivation is often investigated using operant behavioural paradigms such as the Effort for Reward (EfR) task. Performance on these tasks is thought to be influenced by reinforcement learning (RL) mechanisms, such that disorders of motivation can be described in terms of altered interactions with RL processes. Recently, foraging behaviour has been increasingly adopted as an ethologically valid approach to investigating cognitive mechanisms, including motivation. One example of this is the recently developed Effort Based Foraging task (EBF). Foraging behaviour, unlike the strictly controlled and contrived operant behavioural paradigms, involves a series of complex behavioural processes, each potentially driven by different cognitive processes. It is therefore important to identify which portion of the foraging behavioural sequence is driven by the same RL mechanisms as classical operant behavioural paradigms, and therefore can be used to investigate motivation. In this work we set out to establish whether the same RL mechanisms could be used to account for behaviours observed in both EfR and EBF tasks. We identified where, within the EBF task, RL mechanisms were no long sufficient and developed a novel hybrid model of the behaviour. This model successfully accounted for external influences on motivation, including previously unpredicted effects of clinically used dopaminergic drugs. This work reveals that motivation in complex, naturalistic tasks cannot be fully explained by learning-based models alone. Incorporating intrinsic behavioural drives may be needed as neuroscience moves toward more ethological behavioural assays.

neuroscience↗

bioRxiv2025

A Computational Model of Action Specification in the Basal Ganglia

The basal ganglia has traditionally been modelled as a system that represents the available, discrete action space using a finite set of distinct and non-overlapping representational units. This limits these models from addressing questions around how the basal ganglia is involved in action specification: the selection of continuously valued action dynamics such as speed and vigor. In this article we present a novel computational model of the basal ganglia which incorporates vector-symbolic algebras in order to represent continuous action spaces. The stages of model development are presented along with simulation experiments to test the basic properties of the model. This work represents a promising foundational step in providing a mechanistic account for neuroscientific and behavioural evidence implicating the basal ganglia in action specification, thereby filling a gap in the literature.

neuroscience↗