CS Wednesday Talks Fall 2010/11

Fall 2010

September 15: Mara Breen, adjunct assistant professor of psychology
Stress matters: The role of word stress in silent reading
Abstract: We know that spoken language preceded written language historically, and that children learn to read aloud before they read silently. An open question in research on silent reading is the extent to which it mirrors the process of reading aloud. Specifically, are silent readers activating the same sound representations during silent reading that are activated when reading aloud? In this talk, I present data from two eye-tracking studies that provide evidence that readers sound out the stress patterns of words even when reading silently.

September 22: Joanna Morris, associate professor of cognitive science
Morphological priming of regular and irregular verbs
Abstract: In this talk I will present evidence about how the brain recognizes complex words. Do we decompose complex words into their component parts in order to recognize them and reassemble them in production? Or are they stored intact in our mental dictionary? Recent masked priming experiments have brought to light a morphological level of analysis that is exclusively based on the orthographic appearance of words, so that it breaks down corner into corn- and -er, as well as dealer into deal- and -er. Because this level of analysis ignores meaning, it cannot capture the morphological relationship between irregularly inflected words and their base forms (e.g., fell-fall, bought-buy). I will present behavioral and electrophysiological data from an experiment comparing irregular and regular inflections that provide support for a level of morphological analysis that takes semantic as well as orthographic information into account.

September 29: Tony McCaffrey, University of Massachusetts graduate student; Lee Spector, professor of computer science, Hampshire College
A New Psychology of Innovation Based On Two Principles: The Obscure Features Hypothesis and the Just Outside the Box Hypothesis
Abstract: Insight problems (that often involve aha moments) have been studied since the early part of the 20th century. Recent analysis has determined that insight problems are solved by noticing a rarely noticed or never-before noticed (i.e., obscure) feature of the problem's elements and then building a solution based on that obscure feature. Besides toy problems in psychology labs, obscure features are also often crucial for solving engineering design problems (e.g., design a new type of shovel) as the novel designs are usually built upon an obscure feature of the object. This Obscure Features Hypothesis for innovation opens up a research program based on two questions. What inhibits people from noticing the obscure? What techniques can help overcome these sources of inhibition? First, I will present results from my lab on the techniques that have been tested. Second, I will present results from analyzing the structural properties of semantic networks that are important for problem solving. For example, drawing a box around the semantic network of all the common associations for an object shows the common thoughts close to that object, which are "inside the box." The key information for solving a problem is most often just outside the box, literally one or two steps away. This and other structural properties of a semantic network make possible a computer program that can assist humans in problem solving by guiding humans to the parts of the semantic network with the highest probability of being helpful. After winning prize money for this idea at the Spring 2010 UMass Innovation Challenge, this software is being developed and will be beta-tested at Raytheon, Yankee Candle, and Saint-Gobain later this year.

October 6: Ray Coppinger, professor emeritus of biology
Dogs are Lousy Mothers
Abstract: The canids have some of the most complex parental behaviors in the animal world. How could it be possible that dogs (the most recent form  in the genus Canis) have evolved a minimal care pattern, more crude  than that of any other mammal? At first glance it seems unlikely that bad mothering or fathering or even sistering could be a selective advantage--but it seems be true.

October 13: Daniel Altshuler, Mellon postdoctoral fellow in language, mind, and culture
Abstract: One reason why linguists study aspectual markers (e.g. progressive and the perfect) is that they reveal what event parts are linguistically relevant and how these parts are located in time in a given discourse or story. Aspectual markers, however, are rarely precise about the temporal location of a given event part. Inferences about an event's temporal location are also dependent on independently motivated temporal constraints imposed by so-called coherence relations, which characterize the possible ways in which successive utterances could be connected to form coherent discourse. In this talk, I will discuss how aspectual meaning interacts with coherence relations by looking at the Russian imperfective. In this way, I will also provide a sneak preview for my class next semester, Literature and Cognition.

October 20: Jane Couperus, assistant professor of cognitive neuroscience
Evidence for a Two-process Model of Visual Selective Attention
Abstract: Selective attention modulates activity at early levels of visual processing, as is reflected in changes in the P1 event-related potential (ERP) component. Although some have suggested that the process of selection involves primarily signal enhancement (e.g., Mangun et al., 1991), others have suggested that it involves both the enhancement of the signal of the attended stimulus as well as suppression of the unattended stimulus (e.g. Awh, Matsukura, and Serences 2003; Couperus and Mangun, in press, Dell'Acqua et al., 2007). In addition to presenting recent electrophysiological work with adults that begins to provide evidence of these two process, this talk will also examine behavioral and preliminary electrophysiological evidence from children.

October 27: A panel of CS faculty and students, introduced by Laura Wenk, associate professor of education and cognition
What Could Community Engaged Learning Mean in CS?
Abstract: Hampshire College is in the middle of a shift from a Division II Community Service requirement to a Community Engaged Learning (CEL) requirement. These are very different things with big implications for course development and for student work. Come hear about what this requirement means and how some CS students have been meeting the requirement. Perhaps most importantly, help us think about what opportunities we could create to help students mesh their academic learning with the world beyond the classroom. There will be short talks by a few CS students and faculty, followed by lively discussion. Oh, yeah, and of course, there will be pizza, salad, fruit, and the like.

November 3: Joanna Morris, associate professor of psycholinguistics and cognitive science
How wlel do we raed copmlex wrods with trnaspsoed lteters?
Abstract: Recent studies have shown that semantically opaque pseudo-complex words (e.g. corner) prime their embedded pseudo-stems (i.e. corn) as much as semantically transparent word-stem pairs do (e.g. hunter-hunt).  These data suggest suggest that morphological structure affects visual word recognition uniquely via sub-lexical morpho-orthographic segmentation. Here, we show that priming with transparent and opaque primes is differentially affected by transposing morpheme boundary letters. Whereas priming remains intact for transparent items (relative to replaced letter primes; viewer-view ~ vieewr-view < vieakr-view), it is reduced to the level of replaced letter primes for opaque items (corner-corn < corenr-corn ~ coratr-corn). This pattern of results was predicted by a model of complex word recognition that involves both sub-lexical morpho-orthographic and supra-lexical morpho-semantic processing. Transposed letters at the morpheme boundary are harmful for the fine-grained sub-lexical orthographic code that drives morpho-orthographic processing, but not for the coarse-grained lexical orthographic activation that initiates morpho-semantic processing.  

November 10: Daniel Asia, ('71 Alum)

The Act Thereof, and a Few Results
Abstract: The creative process is rich with complexity and fraught with problems. I will speak about some of the questions to ask to help you develop your own best practices in dealing with yours. This will pertain to composers in particular, artists more generally, those engaged in creative thinking, or those who wish to understand more about the creative/cognitive process. This will be supported with the playing of a few of the composer's selected works.

Daniel Asia ('71 Alum) has been an eclectic and unique composer from the start. He has enjoyed the usual grants from Meet the Composer; a UK Fulbright award; Guggeneheim Fellowship; MacDowell and Tanglewood fellowships; ASCAP and BMI prizes; Copland Fund grants; and numerous others. He was recently honored with a Music Academy Award for the American Academy of Arts and Letters. From 1991-1994 he was composer in residence of the Phoenix Symphony. Under a Barlow Endowment grant, he recently finished a new work for The Czech Nonet, the longest continuously performing chamber ensemble on the planet, founded in 1924. He is currently finishing his first opera, The Tin Angel, based on the novel by Paul Pines. The recorded works of Daniel Asia may be heard on the labels of Summit, New World, Attacca, Albany, Babel, and Mushkatweek. For further information, visit the Daniel Asia website at www.danielasia.net. Asia is currently professor of music and director of the Center for the Study of American Ideals and Culture at the University of Arizona.

November 17: Carol Trosset, director of institutional research, "Welsh Concepts of Personhood."

December 1: Lightning Talks

Spring 2011

April 13: Chris Bishop, adjunct assistant professor of computer animation

"Behind the scenes of "Caldera": A computer animated short film production at Hampshire College"
Beginning in the summer of 2009, fellow alum Evan Viera and I have been in production on a new short film titled "Caldera". The project is one of two films being produced through the Bit Films Internships in the School of Cognitive Science, a program that brings together students and professionals dedicated to collaborative, independent graphics work. I will be discussing the interdisciplinary nature of the process behind computer animation and how our program facilitates this process, as well as presenting a showcase of work from "Caldera," which combines the skills of various technical specialists and visual artists.

April 6: Melissa Burch, assistant professor of cognitive development
"Cross-cultural perspectives on autobiographical memory: A comparison of narratives from Chinese, Russian, and English adults"
Socialization processes have been studied extensively within the autobiographical memory literature. The effects of socialization have been explored within the context of parent-child interaction within the U.S. and by comparing parent-child interaction across cultures. It has been hypothesized that different socialization practices reflect the relative importance of social and self identity, with individuality emphasized over group connections for those raised in the U.S. and greater significance of group connections for those raised in China. This perspective has been used to explain the later age of earliest memories for Chinese adults compared to American adults as well as differences in the content of memory narratives. In this talk, I will present data comparing memory ratings and reports from American, Chinese, and Russian adults living in the U.S. Three methods were used to gain information about memory for previous experience. Under some conditions, we replicate findings reported in the literature. When we prompted participants to provide specific types of memory (using a cue word technique, or requesting memories for events related to "shame" and "honor," however, the differences were more limited. I will discuss how these findings may reflect larger cultural values as well as similarities across cultural groups in memory for personally meaningful events.

March 30: Neil Stillings, professor of psychology
Does Hampshire Need More Examinations, Tests, and Quizzes? A Cognitive Psychological Approach
A notable feature of Hampshire's classroom environment is the rarity of exams, tests, and quizzes in comparison to common practice at other colleges. A number of arguments, based on various combinations of principle and evidence from psychological and educational research, have been generated for the relative infrequency of tests at Hampshire. In this talk I explore one potentially challenging line of evidence. An extensive body of recent research demonstrates that testing enhances student learning. On the one hand this research offers an opportunity to revisit aspects of Hampshire's educational philosophy. On the other hand Hampshire's educational philosophy provides a framework for interpreting the testing effect that differs from the ones commonly seen in popular media coverage of the effect.

March 23: Carol Trosset, director of institutional research
Perceptions of Welshness in Patagonia
In 1865, the Welsh became the first European settlers in southern Argentina, and they remain the dominant ethnic group in the state of Chubut. Some towns retain large populations of Welsh-speakers, and a strong awareness of their Welsh cultural heritage. In 2002, Trosset did ethnographic fieldwork in several of these towns. This talk will present an overview of contemporary Welsh culture in Patagonia, which has maintained close ties to Welsh-speaking Wales. Trosset will discuss the results of her structured interview study, which made it possible to compare Patagonian views of Welshness to views held by Welsh people living in Wales, revealing both similarities and differences.

Carol Trosset has been director of institutional rat Hampshire College since 2004. She holds a B.A. from Carleton College and a Ph.D. in cultural anthropology from the University of Texas at Austin. Her published institutional research areas include student approaches to the discussion of diversity issues, student and faculty expectations of advising relationships, and the nature and validity of the information provided in student end-of-course evaluations. Her academic research focused on aspects of Welsh culture, in Wales and among diaspora populations in Argentina and Australia. Before entering institutional research, she held visiting faculty positions at the University of Arizona, Tulane University, the University of Virginia, Grinnell College, and Beloit College.

March 9: Dr. Ernest Lepore, director of the Rutgers Center for Cognitive Science (RuCCS)
Context and Shared Content
Contextualism looms large over cognitive science, linguistics, and philosophy. Ordinary folk have apparently missed the fact that many of our most puzzling and paradox-ridden expressions are sensitive to their context of use familiar words such as "know", "believes", "truth", "good", and "beauty" turn out to require contextualization for application. Fears of the Liar Paradox, the Paradox of the Heap, reconciling Skepticism with Compatiblism, Moral Relativism, all are alleged to vanish once key expressions are recognized as context-sensitive.

Though there is much observational support for contextualism, I want to draw your attention to the observation that most speakers face no difficulty whatsoever in using many of the alleged context-sensitive words to say (or make) the exact same claim, assertion, etc., across a wide array of contexts. So, on the one hand, for many sentences there is evidence that what their utterances contribute depends on features of their contexts of use; while, at the same time, there is evidence (as I will remind you ) that relevantly distinct utterances of these sentences in distinct contexts express agreement.

March 2:  Lee Spector, professor of computer science
Biologically-Inspired Evolution of Computer Programs: Tag-based Modularity in Genetic Programming
Genetic programming is a computational technique that uses ideas from evolutionary biology--random variation and natural selection--to automatically produce computer programs. Natural adaptive systems invariably make use of modularity, and human programmers are more productive when they use modular program architectures. For these reasons it is generally accepted that genetic programming systems will be more powerful when they can more readily evolve programs with modular structures.

In this talk I will present a new technique for evolving modular programs, based on the use of "tags" that evolving programs can employ to label and refer to code fragments. The essential idea of a tag is that it supports binding through matching, even though specific tags may initially have no intrinsic meaning and even though matches may sometimes be inexact. Everyday examples of tags given by John Holland, who first developed the concept, include banners or flags used by armies and the active sites that enable antibody/antigen binding. Systems based on tags have been used to explore a variety of phenomena including the evolution of altruism, but the work that I will present here is the first to apply the concept to general program evolution. I will demonstrate that tag-based modules readily evolve, that this allows problem solving effort to scale well with problem size, and that the technique is effective even in complex environments for which previous techniques perform poorly.

This is joint work with Hampshire College student Brian Martin, Hampshire College alum; Brandeis University graduate student Kyle Harrington; and University of Massachussetts Amherst graduate student Thomas Helmuth.

February 23: Charles Ross, assistant professor of evolutionary biology
The personal nature of speciation
Speciation is often considered as a population level phenomenon described in terms of divergence of genes and genomes across groups. For example, populations may diverge in their genetic makeups due to geographic isolation. The divergence of specific genes may lead to reproductive isolation across these populations, resulting in speciation. These perspectives--looking from the viewpoint of both populations and genes--are informative for understanding the pattern and processes of speciation. Speciation from the perspective of individuals is often overlooked even though individuals are the functional components that make up populations as well as the "phenotypic results" of gene expression (along with environmental influences). Looking at speciation from the perspective of individuals can help to understand the mechanics how populations are isolated, how genes contribute to reproductive isolation, and how speciation actually happens "on the ground."

Charles L. Ross, assistant professor of evolutionary biology, received his B.S. and M.S. in biology from Stanford University, and his Ph.D. in ecology and evolutionary biology from Cornell University. He did postdoctoral work at the University of Arizona and New Mexico State University. Charles studies the ecological and evolutionary genetics of hybrid zones and speciation, specifically in crickets. His research and teaching interests include all aspects of evolutionary biology, as well as population genetics, molecular ecology, entomology, and genomics. Other interests include Ultimate, backpacking, and good wine.

February 16: Paul Dickson, visiting assistant professor of computer science
Ubiquitous Computer:
Over the years computers have shifted from being room-size behemoths that only a limited number of people use to hand-held devices that almost everyone has. This ubiquitousness of computers changes their perceived uses and interfaces. This talk will address questions of where computing is going and what computers will be used for in the future. We will look at how mobile computers are effecting human interaction and changes it is having on society. This talk will also address what the possibilities are at Hampshire for being a part of this change.

Paul Dickson, visiting assistant professor of computer science, received a Ph.D. and an M.S. in computer science from the University of Massachusetts Amherst and a B.S. in engineering from Swarthmore College. His main interests are computer vision and education technology. He is currently working on projects that combine these areas and use image-processing techniques to capture automatically and index the classroom experience.



Stay In Touch
With School of Cognitive Science (CS)
Snail Mail
School of Cognitive Science
Hampshire College
893 West Street
Amherst, MA 01002
Phone + Email
Fax: 413.559.5438