[Paper Review] Reaction time impairments in decision-making networks as a diagnostic marker for traumatic brain injuries and neurodegenerative diseases

Pedro D. Maia, J. Nathan Kutz|arXiv (Cornell University)|Aug 20, 2016

Neural dynamics and brain function81 references3 citations

TL;DR

This study proposes a computational model linking focal axonal swellings (FAS)—a cellular hallmark of traumatic brain injury (TBI) and neurodegenerative diseases—to impaired decision-making via disrupted spike-train encoding in excitatory-inhibitory neural networks. Using population codes and decision variable integration, the model demonstrates that reaction time deficits, not accuracy loss, are the primary indicator of network-level damage, offering a non-invasive diagnostic framework using simple cognitive or motor tasks to assess injury severity and disease progression.

ABSTRACT

The presence of diffuse Focal Axonal Swellings (FAS) is a hallmark cellular feature in many neurodegenerative diseases and traumatic brain injury. Among other things, the FAS have a significant impact on spike-train encodings that propagate through the affected neurons, leading to compromised signal processing on a neuronal network level. This work merges, for the first time, three fields of study: (i) signal processing in excitatory-inhibitory (EI) networks of neurons via population codes, (ii) decision-making theory driven by the production of evidence from stimulus, and (iii) compromised spike-train propagation through FAS. As such, we demonstrate a mathematical architecture capable of characterizing compromised decision-making driven by cellular mechanisms. The computational model also leads to several novel predictions and diagnostics for understanding injury level and cognitive deficits, including a key finding that decision-making reaction times, rather than accuracy, are indicative of network level damage. The results have a number of translational implications, including that the level of network damage can be characterized by the reaction times in simple cognitive and motor tests.

Motivation & Objective

• To establish a mechanistic link between cellular-level FAS and network-level decision-making impairments in neural circuits.
• To address the diagnostic gap in detecting early-stage neurodegenerative and traumatic brain injuries by identifying measurable behavioral signatures.
• To develop a computational framework that translates FAS-induced spike-train distortions into interpretable psychophysical deficits in decision-making tasks.
• To demonstrate that reaction time, not accuracy, is the most sensitive indicator of network-level damage from FAS.

Proposed method

• Modeling a biologically plausible excitatory-inhibitory (EI) neural network that produces low-dimensional population codes in response to stimuli.
• Mapping neural activity trajectories to a reduced-dimensional space via principal component analysis (PCA) to represent evidence accumulation.
• Incorporating FAS as virtual lesions that distort spike-train propagation, leading to anomalous evidence production (e.g., biased, confused, or amplified signals).
• Simulating decision-making via a drift-diffusion-like accumulation model where evidence is integrated over time until a decision boundary is reached.
• Quantifying network-level damage through metrics such as transmitted information (TI) decay and reaction time shifts under varying FAS distributions.
• Using experimental data from Wang et al. (2011) to inform realistic FAS distributions and their impact on signal transmission and decision performance.

Experimental results

Research questions

• RQ1 How do focal axonal swellings (FAS) disrupt spike-train encoding and propagation in decision-making neural networks?
• RQ2 To what extent do FAS-induced distortions in evidence production affect decision accuracy and reaction time in binary choice tasks?
• RQ3 Can reaction time deficits serve as a more sensitive diagnostic marker than accuracy loss in detecting network-level damage from FAS?
• RQ4 How does the distribution and severity of FAS correlate with measurable changes in psychophysical performance curves (accuracy vs. speed)?
• RQ5 Can simple cognitive or motor tasks be used to non-invasively assess the level of FAS-related network damage in TBI and neurodegenerative diseases?

Key findings

• Reaction time impairments increase proportionally with the level of network injury caused by FAS, making it the most sensitive indicator of damage.
• Even with correct responses, reaction times are significantly slowed in the presence of FAS, indicating that accuracy alone is insufficient for diagnosing subtle neural network dysfunction.
• FAS lead to anomalous evidence production, including biased errors, confusion between evidence types, and amplification of small or false signals.
• Transmitted information (TI) in the network decays with increasing FAS injury, and this decay is more pronounced under heterogeneous FAS distributions compared to homogeneous ones.
• The model predicts that reaction time deficits can be reliably measured using tablet-based cognitive or motor tasks, enabling non-invasive, repeatable monitoring of disease progression or TBI recovery.
• The framework supports the use of baseline reaction time testing in athletes before matches to detect post-injury slowing, even when accuracy remains intact.