The Science of Intelligence

The problem of human intelligence — its nature, how the brain produces it, and how it could be replicated in machines — is a deep and fundamental problem that cuts across multiple scientific disciplines. Philosophers have studied intelligence for centuries, but it is only in the last several decades — and in particular in the last 3 years — that developments in science and engineering have made questions such as these approachable: How does the mind process sensory information to produce intelligent behavior, and how can we design intelligent computer algorithms that behave similarly? What is the structure and form of human knowledge — how is it stored, represented, and organized? How do human minds arise through processes of evolution, development, and learning? How are the domains of language, perception, social cognition, planning, and motor control combined and integrated? Are there common principles of learning, prediction, decision, or planning that span across these domains?

This class provides instruction on the mechanistic basis of intelligence, addressing the questions of how the brain produces intelligent behavior and how we are beginning to replicate aspects of human intelligence in machines. There are now intelligent machines  that pass reasonable forms of the Turing test. How does their intelligence compare to human intelligence? Is something still missing? If yes, what? We will focus on a quantitative computational approach, combining experimental techniques in neuroscience and cognitive science with computational modeling to elucidate the computational architecture of artificial and natural intelligence and how much it depends on the properties and limitations of the physics and biophysics of the computational mechanisms such as neurons, synapses and transistors.

Through lectures by the various members of the Center for Brains, Minds, and Machines (CBMM), the Quest, BCS and EECS the course will explore recent progress in  understanding the problem of intelligence. Ironically, our understanding of intelligent and biological machines is significantly lagging the development of impressive applications. The lectures include empirical studies not only of artificial systems but also of  humans and primates using psychophysical, imaging, and physiological tools.

For students:

The course has two components, both in terms of content and evaluation:

1) Weekly lectures by researchers on different topics within the science of intelligence — attendance / participation in lectures is required, and each lecture will have a short online quiz component (see below).

2) A semester-long research project (see below).

Each student will be assigned a week during which they are responsible for preparing a short (3-5 question, true/false and multiple choice) quiz on Gradescope based on the researcher lecture of that week. The quiz should be written in such a way that for anyone who attended and paid attention the lecture, it is easy and takes at most a few minutes to complete. Course staff will write the quiz for the very first lecture to serve as an example; the goal of the quiz is to serve as a sanity check, recapping some of the most basic and important ideas from the lecture.

For the research project, project ideas will be presented early on in the year. Students will, independently or in teams of 2, work on their projects for the rest of the semester with moderate supervision/assistance from their mentor. Mentors include the course instructors, TAs, as well as PhD students and postdocs in other labs who have come up with research project proposals and have volunteered to mentor students in teh course.

As this is a CI-M class, students will be expected to present their projects at the end of the semester. Instruction and practice in oral and written communication will be provided. The class is a great opportunity for students who plan to apply to graduate school and/or pursue a career in research, since it is important in both cases to be able to communicate research effectively both orally and in writing. Classes will be held in person unless otherwise specified.

Canvas Website: https://canvas.mit.edu/courses/28355