Understanding Human Thought: A Computational Approach

Attention Modeling

Selective attention is a fundamental process where individuals focus on specific stimuli while filtering out others. In our simulator, we can manipulate attentional resources by altering stimulus salience and introducing competing demands.

A simplified model of attention involves a ‘gain’ factor applied to each sensory input. Higher gain values prioritize stronger signals, mimicking the brain's ability to amplify relevant information.

Gain = Stimulus Strength / Noise Level

Memory Systems Simulation

The human memory system is often divided into sensory memory, short-term memory (STM), and long-term memory (LTM). We can simulate these stages to observe how information is encoded, stored, and retrieved.

STM capacity is limited; our simulation demonstrates this by tracking the number of items a subject can hold in working memory before decay sets in. This highlights the role of rehearsal.

STM Capacity ≈ 7 ± 2 Items

Decision-Making Processes

Cognitive models of decision-making often incorporate biases and heuristics, which are mental shortcuts that simplify complex choices. We can simulate these biases within the simulator.

A basic prospect theory model uses loss aversion to determine an individual’s response to risk. The greater the potential loss, the more sensitive a subject will be to a given outcome.

Utility = -w * (Loss or Gain)

Network Dynamics & Cognitive Load

Cognitive processes are often represented as networks of interconnected nodes. Changes in connectivity and activity patterns can be simulated to understand how information flows within the brain.

Introducing a high cognitive load – for example, by increasing the complexity of a task – will disrupt network dynamics and potentially lead to errors. This demonstrates the limitations of working memory.

Cognitive Load = Number of Concurrent Processes * Difficulty Level

Frequently asked questions

Try it live

Everything above runs in your browser — open SPH Fluid and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.