Laboratory for Brain and Machine Intelligence (BML)
Artificial intelligence intersects with neuroscience
Our question : how do brains and machines learn to predict the future?
There is nearly 50 years of evidence to suggest that the brain has multiple separate modes of learning and inference about the world, each of which can guide behavior in unique ways:
- model-free and model-based reinforcement learning
- incremental and one-shot inference
- planning and cognitive control
After decades of study, we begin to understand how these learning systems interact with each other in order to ultimately produce coherent behavior. Addressing this question is crucial for understanding why the balance between those learning systems might sometimes break down in learning disorders, addiction and psychiatric disease, in which people often fail to suppress inappropriate behaviors in spite of the fact that such behavior ultimately leads to highly adverse consequences.
A central goal of both artificial intelligence and cognitive neuroscience is to understand human cognitive processes that are flexible enough to perform a wide range of tasks. In this regard, it is becoming widely recognized that hierarchical control of learning systems may be both the way the human brain actually works and the optimal design for an artificial intelligence that operates under constraints on performance, time, and energy.
Our goal : we aim to understand learning, inference, and cognitive control on the deepest level.
Recent studies have investigated neural mechanisms of different types of learning through a combination of various techniques measuring brain activity with computational learning models. However, little is known about how the brain determines which of these sub systems guides behavior at one moment in time. Our research interests are to develop a neurocomputational theory of how the brain, arguably at the higher level in the cognitive hierarchy of the prefrontal cortex, allocates control over behavior to multiple types of brain's subsystems for learning and inference.
We believe this understanding opens the possibility for making scientific and technological advances.
This theory will enable us to
(1-Neuroscience) understand why a breakdown of the arbitration control occurs in psychiatric disorders,
(2-Bioengineering) develop neuromorphic algorithms for restoring stability to prefrontal cortex,
(3-Artificial Intelligence) design brain-inspired AI systems that surpass the performance of humans, and ultimately
(4-Human Intelligence) appreciate human intelligence that possesses remarkable abilities to deal with trade-offs between performance, energy, and time.