title: The neural basis for uncertainty processing in hierarchical decision making
abstract: Hierarchical decisions in natural environments require processing uncertainty across multiple levels, but existing models struggle to explain how animals perform flexible, goal-directed behaviors under such conditions. Here we introduce CogLinks, biologically grounded neural architectures that combine corticostriatal circuits for reinforcement learning and frontal thalamocortical networks for executive control. Through mathematical analysis and targeted lesion, we show that these systems specialize in different forms of uncertainty, and their interaction supports hierarchical decisions by regulating efficient exploration, and strategy switching. We apply CogLinks to a computational psychiatry problem, linking neural dysfunction in schizophrenia to atypical reasoning patterns in decision making. Overall, CogLink fills an important gap in the computational landscape, providing a bridge from neural substrates to higher cognition.
To me the paper is still very interesting, but concerns about computationally intractability and the hardness of approximation questions made me dig deeper.
> Specifically, our model aims to bridge biological circuits and computations of uncertainty in a tractable manner.
> To address this, we have carefully reframed our claims throughout the manuscript to emphasize that the model is a hypothesis generator rather than a definitive representation of biological circuitry.
Under the "All models are wrong, some are useful" I have no doubt this will be useful to some. But I will admit that their claims in the response to Reviewer Comment 4.5 that they "emphasize that the model is a hypothesis generator" doesn't match the published paper IMHO; and that negatively impacted my view of the claims in an admittedly probably unfair manner.
[0] https://static-content.springer.com/esm/art%3A10.1038%2Fs414...
“Inside the Matrix: How the Brain Rewrites Reality When Learning Goes Wrong”
“Neural Jedi Training: How the Brain Masters Skills—and Sometimes Turns to the Dark Side”
“Glitch in the Mind: Why Our Brain’s Learning System Sometimes Crashes”
“Braincraft: How We Level Up Knowledge—and Why We Sometimes Rage Quit”
“Neural Flight School: Why the Brain Sometimes Loses Control of the Plane”
“Mind Avengers: When Brain Circuits Team Up—and When They Fight Back”
“Stranger Thoughts: What Brain Simulators Reveal About Learning Gone Weird”
“Control+Alt+Brain: Rebooting How We Learn and Unlearn”
“Mind Simulator 2.0: Debugging the Brain’s Learning Software”
“Fast & Curious: How the Brain Learns at High Speed—Until It Skids Out”
"A Cut Above - How the Prefrontal Cortex Dominates the Striatum Like a High-End Hair Salon"
There is something about a hidden power in our minds to reinvent themselves into capable instruments of our own wills.
What these substance craving minds think of as reality is in actuality their thoughts feelings and beliefs.
Undeceiving the self rewrites reality with a sensibly competent version of one’s own self as an actor.
You prompted chatgpt to create movie based titles and then passed it off as regular output.
With the prompt "suggest 10 titles for this article" + the article text, I get the following _normal_ titles.
1. When the Brain Misreads the World: How Uncertainty Shapes Thought and Behavior
2. CogLinks: A Virtual Brain That Teaches Us How the Mind Adapts
3. The Neural Balancing Act: How the Brain Decides Under Uncertainty
4. Modeling Mental Flexibility: Simulating How the Brain Learns and Adapts
5. Inside the Decision Machine: How New Models Reveal the Brain’s Hidden Algorithms
6. Uncertainty, Meaning, and Misfires: Understanding the Neural Roots of Psychiatric Disorders
7. When Circuits Go Off Course: What a Virtual Brain Teaches Us About Mental Illness
8. The Thalamic Switchboard: Linking Flexibility and Habit in the Human Mind
9. From Neurons to Algorithms: Building a Bridge Between Brain Biology and Psychiatry
10. Toward Algorithmic Psychiatry: Simulating Brain Circuits to Decode Mental Disorders
Realized that the article has nothing to do with planes -->
Closed the article
jt2190•11h ago
IAmBroom•10h ago