Since I played it the first time when I was young, Black Box has been one of my favorite games. I think that studying cognitive psychology and psychology of language is like playing the Black Box game in the sense that we investigate the structure of human mind by presenting stimuli and observing responses. Fascinated by how well-designed experiments could tell us about apparently "unobservable" things, I decided to study psychology of language and cognition.
I am interested in the dynamics of the human cognitive system both on a slow time scale (learning dynamics) and on a fast time scale (processing dynamics). More specifically, I study how people develop symbolic knowledge (e.g., categories, rules, grammar) from experience and how they compute the representation of structured symbolic knowledge (e.g., syntactic tree structures) online. My work heavily relies on the use of the continuous representation space, which results in interesting arguments for continuity between finite state systems and recursive systems, the gradient between local coherence and garden path phenomena, and dynamic encoding of structural uncertainty.
My general interest in the (neurally plausible) representation and processing dynamics of structured symbolic knowledge focuses on one central research question: How do we deal with local ambiguity? Information is distributed in time and space but the capacity of our perceptual system and memory is highly limited. Thus, we must sequentially collect and integrate information to build up a globally coherent interpretation. The problem is that partial information available at a given point is typically ambiguous (local ambiguity), creating serious computational challenges for the human cognitive system.
To answer my research questions, I use behavioral experiments (e.g., self-paced reading, artificial language learning, visual world paradigm eye-tracking studies) to investigate human behaviors and try to explain them by constructing and investigating computational models (usually, recurrent neural network models). I use the conceptual and computational tools of dynamical systems theory to understand both human and model behaviors.
"I have been working hard on it [Ulysses] all day," said Joyce.
"Does that mean that you have written a great deal?" I said.
"Two sentences," said Joyce.
I looked sideways but Joyce was not smiling. I thought of Flaubert.
"You have been seeking the mot juste?" I said.
"No," said Joyce. "I have the words already.
What I am seeking is the perfect order of words in the sentence.
There is an order in every way appropriate. I think I have it."
--- Frank Budgen, James Joyce and the Making of "Ulysses"
Last updated on 2017-11-15