Cognitive Science Research

Fred Adams (linguistics and cognitive science) is currently investigating the theory known as embodied cognition. This theory has many claims, but the one he is most interested in is the claim that cognitive processing is going on in sensory and motor areas of the brain. The traditional view of cognition is that cognition only happens after perceptual input and before motor output. Many advocates of embodied cognition claim that the traditional view is false and that cognitive processing takes place across the sensory motor divide. More extreme views go so far as to say cognitive processing can go all along the perception-action loop, and even extend beyond the boundaries of the body and brain. Adams has written several articles on embodied cognition and has presented papers on the topic in Berlin, Brazil, Switzerland and Australia.

Recent research has revealed a robust and surprising effect of emotion on perception: when people see a task-irrelevant emotional picture embedded within a rapid stream of scene pictures their attention is so preoccupied by the emotional stimulus that they spontaneously experience a brief period of functional "blindness", in that for about 0.8 sec. they fail to notice a target that they are actively searching for within that stream, an effect known as emotion-induced blindness or EIB.

This attentional competition between the emotional distractor and the target during EIB could be occurring in several different brain areas, ranging from early perceptual stages to late stages involving short-term memory. James Hoffman (psychological and brain sciences) explores this issue using event-related brain potentials (ERPs) which measure the electrical activity of the brain using sensors located on the scalp, and has found that both the irrelevant emotional picture and the relevant target produced a negative component over visual brain areas (occipital temporal cortex). This component, known as the N2, reflects visual attention being allocated to a stimulus: in one case, involuntarily (the negative picture) and in the other, voluntarily (the target picture), and is profoundly affected by EIB.

An interdisciplinary team from linguistics and mechanical engineering is building bridges between language acquisition and robotics using formal methods

An interdisciplinary research group at UD is working on how robots can be more autonomous and responsive to their environments, as well as cooperate with each other in the field. Led by Jeffrey Heinz (linguistics and cognitive science) and Bert Tanner (mechanical engineering), The group is investigating the application of novel grammatical inference methods to problems of controlling cooperative as well as competitive interactions between robots and other dynamical systems.Grammatical inference is integrated in control design as a tool for learning unknown system behavior, and subsequently refining control strategies.

Their National Science Foundation (NSF)-funded project focused on improving programming codes for increasingly complex robot capabilities, by using principles from human language learning. [Click here to view an an NSF video and article about their project.]

A U-shaped curve in a cognitive-developmental trajectory refers to a three-step process: good performance followed by bad performance followed by good performance once again. U-shaped curves have been observed in a wide variety of cognitive-developmental and learning contexts. U-shaped learning seems to contradict the idea that learning is amonotonic, cumulative process and thus constitutes a challenge for competing theories of cognitive development and learning. U-shaped behavior in language learning (in particular in learning English past tense) has become a central topic in the Cognitive Science debate about learning models. Antagonist models (e.g., connectionism vs. nativism) are often judged on their ability of modeling or accounting for U-shaped behavior.

The prior literature is mostly occupied with explaining how U-shaped behavior occurs. Instead, John Case (computer science) is interested in the necessity of this kind of apparently inefficient strategy. His group reviews a body of results in the abstract mathematical setting of (extensions of) Gold-style computational learning theory addressing a mathematically precise version of the following question: Are there learning tasks that require U-shaped behavior? All notions considered are learning in the limit from positive data. The pattern emerges that, for parameterized, cognitively relevant learning criteria, beyond very few initial parameter values, U-shapes are necessary for full learning power.