Funded Grants

Self-directed learning: Understanding the interactions between decision making, learning, and memory

Hayy Aben Yaqzan ("Alive Son of the Vigilant”) is often cited as the first philosophical treatise or novel presented in literary form. Written in 1105 by the Islamic novelist, theologian and physician Abubekar Ibn Tufayl, it tells a story of an isolated feral child — an autodidatic prodigy — who without parents, teachers, or language discovers the laws of nature through self-guided exploration and experimentation (Ibn Tufayl, 1105/1996). The pedagogical philosophy first introduced by Ibn Tufayl emphasizes our human capacities for curiosity, selective information sampling, and inference, ideas which continue to reverberate through contemporary educational policy and psychological research. My research explores essentially the same topic viewed almost 900 years later through the lens of another medieval Islamic invention, the algorithm.

In particular, I am interested in how people bootstrap understanding from their self-guided interactions with the world around them. For example, how are we so good at figuring out how something works by tinkering with it? How do we formulate questions with the goal of gaining knowledge and reducing our uncertainty? How do our choices to gather information affect our memory or conceptual knowledge? Such questions strike at the heart of what makes us such an adaptable and intelligent species.

A central goal of my work is to convert these broad questions about human learning into precise scientific hypotheses. As a former engineer, I have always drawn inspiration from the famous Richard Feynman quote "What I cannot create, I do not understand. " This idea—that understanding can be expressed by building an equivalent model system—is central to both my research and teaching philosophies. In particular, my research approach is organized around the construction of computational models of human cognition. In psychology, computational models are simply psychological theories that have been specified in enough detail to be run as computer programs. Modeling requires theorists to be precise and explicit, aiding communication between scientists and enhancing scientific rigor.

Many of the modeling ideas exposed in my work draw inspiration from machine learning research, which aims to develop artificial systems with capabilities similar to humans. While machine learning researchers often look to the capacities of human intelligence for inspiration, a premise in my work is that this dialog can also run productively in the other direction. By comparing the behavior of people to intelligent algorithms that can plan, reason, decide, and learn in complex domains, we can better understand what makes humans so uniquely smart.

With support of the James S. McDonnell Foundation, my future plans will evolve along two related fronts. First, I plan to continue to expand my research on self-directed learning, specifically focusing on how selective, self-directed interactions can shape people’s understanding of the world. Second, I plan to develop an open-source software infrastructure that will reduce the redundant effort expended by cognitive scientists implementing (and replicating) behavioral experiments.