The Beckman Institute Graduate Student Seminar Series presents the work of outstanding graduate students working in Beckman research groups. The seminars begin at Noon in Beckman Institute Room 1005 and are open to the public. Lunch will be served.
Aging, Parafoveal Preview, and Semantic Integration in Sentence Processing: Testing the Cognitive Workload of Wrap-up
Brennan Payne, Human Perception and Performance
Consistent with the idea that aging has its largest negative impact on fluid abilities, which place heavy demands on attentional processes, age deficits become more pronounced for aspects of language comprehension that are highly effortful. For example, a number of studies suggest that the ability to retain message-level semantics show age-related declines. However, the mechanisms underlying these effects are not well understood. One potential mechanism associated with online integrative semantic processing during reading is a phenomenon called wrap-up, characterized by peaks in processing time at clause and sentence boundaries. In this talk, I present an eye-tracking experiment that tests the degree to which wrap-up demands attentional resources by examining its effects on the parafoveal preview benefit (PPB) in younger (18-29) and older (61-84) adults. The PPB is defined as the facilitation in processing word N+1 based on information extracted while the eyes are fixated on word N, and is known to be reduced by processing difficulty at word N. Findings indicated that wrap-up reduced the PPB similarly for both younger and older adults in early processing measures. However, for later-pass measures, sentence wrap-up differentially reduced the PPB of word N+1 among older adults. Collectively, these findings suggest that the integration of information across sentences is cognitively demanding and may be less efficient with advancing age.
Democracy and Adaptation Effects of India's Rural Employment Guarantee Scheme
Harry Fischer, Social Dimensions of Environmental Policy
Countries around the world are gearing up to respond to anticipated climate change. The rural poor of developing countries are especially vulnerable, due to their high dependence upon their local environment for survival. The uncertainty of future climate conditions at the local level means that now, more than ever, rural development efforts must be responsive to citizens' changing needs. Programs may be most effective when they foster democratic participation to incorporate citizens' needs and aspirations into local assistance efforts. Yet, past experience with participatory development initiatives has shown that effective participation is often not forthcoming, while development outcomes are often highly inequitable. Moreover, little is actually known about factors that enable communities to channel development projects towards addressing climate-related vulnerabilities.
A recent policy in India, The National Rural Employment Guarantee Act (NREGA), provides an excellent opportunity to explore these themes. NREGA provides a social safety net through the provision of minimum-wage employment generated through state-funded rural development activities. Projects are ostensibly selected at the local level through democratic deliberation. Through both qualitative analysis and quantitative modeling, I examine NREGA's implantation in the northern Indian state of Himachal Pradesh. More equitable and democratic project implementation, I hypothesize, is the outcome of the dynamic interactions of multiple actors traversing the state-society divide. Access to critical adaptation knowledge and capacity to influence development decisions are likely to be bolstered by homogeneity of endowments and interests, a vibrant civil society, human capital, access to mass media, and a local history of coping with climatic stresses.
Studying the Interaction Between Language and Action Using a Humanoid Robot
Logan Niehaus, Artificial Intelligence
The field of cognitive robotics takes the position that language – and more generally, cognition – is a necessarily embodied phenomenon. Under this view, developmental, social, and biological factors, in addition to the traditional paradigm of data processing, all contribute to our cognitive abilities. Humanoid robots, such as the iCub used by our lab, allow us to employ models which explore how language interacts with the complete range of sensorimotor experiences. Specifically this talk explores how the robot uses its motor skills to ground the various action words it learns, and how this grounding can be used to learn even more complex skills. We present a computational framework of this process based on statistical models, with which the robot is able to learn both words and actions through social interaction with a human instructor. We also demonstrate that once this stage of learning is complete, the robot may use the same models to reproduce the actions it has learned given a verbal command from the instructor.