Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2508.07471

Modeling bias in decision-making attractor networks

Safaan Sadiq
View original source ↗Read source document ↗Export metadata ↓

Abstract

Attractor neural network models of cortical decision-making circuits represent them as dynamical systems in the state space of neural firing rates with the attractors of the network encoding possible decisions. While the attractors of these models are well studied, far less attention is paid to the basins of attraction even though their sizes can be said to encode the biases towards the corresponding decisions. The parameters of an attractor network control both the attractors and the basins of attraction. However, findings in behavioral economics suggest that the framing of a decision-making task can affect preferences even when the same choices are being offered. This suggests that the circuit encodes both choices and biases separately, that preferences can be changed without disrupting the encoding of the choices themselves. In the context of attractor networks, this would mean that the parameters can be adjusted to reshape the basins of attraction without changing the attractors themselves. How can this be realized and how do the parameters shape decision-making biases? In this PhD thesis we study this question mathematically in the context of threshold linear networks, a common firing rate model used in computational neuroscience. (Note: This is an abbreviated abstract. Please see the document for the full abstract.)

Explore related subjects

q-bio.NC

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Safaan Sadiq. 2025-08-10. Modeling bias in decision-making attractor networks. https://arxiv.org/abs/2508.07471

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

ELiSe: Efficient Learning of Sequences in Structured Recurrent Networks

Behavior can be described as a temporal sequence of actions driven by neural activity. To learn complex sequential patterns in neural networks, memories of past activities need to persist on significantly longer timescales than the relaxation times of single-neuron activity. While recurrent networks can produce such long transients, training these networks is a challenge. Learning via error propagation confers models such as FORCE, RTRL or BPTT a significant functional advantage, but at the expense of biological plausibility. While reservoir computing circumvents this issue by learning only the readout weights, it does not scale well with problem complexity. We propose that two prominent structural features of cortical networks can alleviate these issues: the presence of a certain network scaffold at the onset of learning and the existence of dendritic compartments for enhancing neuronal information storage and computation. Our resulting model for Efficient Learning of Sequences (ELiSe) builds on these features to acquire and replay complex non-Markovian spatio-temporal patterns using only local, always-on and phase-free synaptic plasticity. We showcase the capabilities of ELiSe in a mock-up of birdsong learning, and demonstrate its flexibility with respect to parametrization, as well as its robustness to external disturbances.

q-bio.NC↗
arXiv2026

Three Failures of Pain Location: Why Its Diagnostic Utility Is Three Quantities, Not One

Patient-reported pain location is diagnostically decisive for some presentations and nearly uninformative for others. The prevailing account treats this as one gradient of diagnostic utility set by anatomical complexity. That explanation conflates three epistemically distinct failures, each with its own mathematics, its own optimal instrument, and its own public-health consequence. In anatomical multiplexing, many structures share one location: a non-identifiable inverse problem. In delocalized amplification - clinically, central sensitization or nociplastic pain - a centrally driven pain-behaviour pattern replaces the peripheral generator: a change of generative model. In referred and atypical displacement, location is hypothesized to shift in a systematic, person-dependent way: a group-conditional bias whose direct evidence is still open. The three are one Bayesian inference problem failing at different nodes - the likelihood, the model class, and the group-conditional prior - with a fourth node at the report itself. The formal development is in a companion paper; this paper states what each model shows and what follows clinically. Re-examination finds that the published "high-utility" accuracy band leans on overstated specificity (Lipton et al., 2003; Bruyninckx et al., 2008; Devillé et al., 2000), so the gradient is real but flatter than drawn. The well-evidenced finding that better detection alone does not improve outcomes when treatment uptake lags is about the care pathway, not perception - a distinction the three-way split makes visible and a single utility number hides. The paper organizes the failures along a why-location-fails axis, distinct from the nociceptive/neuropathic/nociplastic taxonomy (Kosek et al., 2016), and sets out the study that would test the one prediction still open.

q-bio.NC↗
arXiv2026

A Spiking Neural Network Model of Elementary Self-Consciousness via Endogenous Default Mode Network Dynamics

Understanding the neurobiological mechanisms underlying self-referential cognition and baseline self-consciousness remains a fundamental challenge in computational neuroscience. In this work, we propose a large-scale computational model incorporating a 10,000-neuron spiking neural network (SNN) based on Izhikevich dynamics. The network is structured into two interacting subsystems: a sensory processing layer (5,000 regular-spiking cortical neurons) and an endogenous Default Mode Network (DMN) pacemaker subsystem (5,000 intrinsically bursting neurons). The DMN layer is modulated by continuous tonic currents reflecting ascending brainstem neuromodulation, maintaining intrinsic, autonomous bioelectric rhythms independent of external sensory input. To represent top-down cognitive modulation, synaptic weights are hierarchically structured such that DMN-to-network projections exceed sensory-level connections. Through numerical simulations using a modified two-step Euler integration scheme, we demonstrate how endogenous pacemaker activity interacts with transient external sensory perturbations, providing an elementary mathematical framework for the emergence of a persistent, self-sustaining neural representation of "Self".

q-bio.NC↗