Di Liu

I am a third-year Ph.D. candidate at the Lab for Child Development, Psychological & Brain Sciences department, Johns Hopkins University, advised by Dr. Lisa Feigenson. My research aims to reveal how we use 'surprise' to reason, explore, and learn. 

Surprise signals a knowledge gap and guides what information we seek out next. But the process compresses many mysteries for cognitive scientists: How do we represent what surprised us? How do we decide what information would resolve it? How do we know when we have learned enough to explain it? 

I study these questions in infants (powerful causal learners before formal education) and adults (who have mature knowledge and strategies). By examining how each group represents surprising events, searches for information, and settles on explanations, I aim to characterize how surprise supports learning across development. 

In particular, my research covers topics including: 

• How infants and adults represent objects that surprise them: After someone sees an object violate a basic physical expectation, what happens next in their representation of that object? One possibility is that they stop treating it as an object at all and stop forming predictions about it. Another is that they rapidly update their beliefs, so they actively expect it to continue to behave surprisingly. A third is that they continue to expect it to obey physical principles, but do not specify which principle was violated. A fourth is that they keep track of the specific violation, and use that record to guide what they attend to and learn next (Spoiler: current evidence suggests the fourth! Stay tuned!) 

• Surprise in a social context: As adults, we know that surprise signals encountering new information and results in a change in what someone knows. How does this understanding develop? In this work, I ask how infants (1) use others' surprise to guide their own information seeking, (2) use this emotion to infer the emoter's subsequent actions, and (3) represent the cause of surprise, which is epistemic by nature. Specifically, can they treat surprise as evidence of discovering something that is new for that particular emoter? 

• Surprise in science: I'm also collaborating with Dr. Marina Dubova on a project testing how experts make sense of and gain inspiration from surprising information. 

• The model of curiosity and information gain: Work such as the Ames-Hunter model asks when infants prefer familiar versus novel information. In our work on surprise, we find that infants prefer interacting with the surprising object despite it being familiar. I aim to ask: how do infants decide that they have gathered enough evidence to restore a coherent conceptual model? And does that information gain function differ from the one infants use when they are building a perceptual description of an object?