Hampshire College

CS Wednesday Talks Spring 2009

February 4: Paul Dickson, visiting assistant professor of computer science.
Storing a Lecture: How to record automatically the classroom experience.
Thousands of lectures and presentations are given around the world each day. Many include information that would be useful to access later, whether the lecture is presented in a school classroom, an office training seminar, or any of a number of situations. This talk will focus on Presentations Automatically Organized from Lectures (PAOL), an automatic lecture recording system that attempts to store the complete classroom experience. PAOL captures all presented material and compiles it into a presentation that enables search of lecture content. PAOL exists at the point where educational technology and computer vision meet. PAOL is inexpensive and transparent to the lecturer, making it ideal for recording lectures wherever they may be given.

Paul is a visiting assistant professor of computer science at Hampshire College. His research focuses on educational technology and computer vision. Paul is currently working on continuing development of Presentations Automatically Organized from Lectures (PAOL) a system that combines his research interests.

February 11: Kathryn Lord, assistant professor of animal behavior.
Interspecies socialization in dogs and wolves, or why wolves just aren’t that into you.
Dogs and wolves are members of the same species, but they are quite obviously different in their behaviors towards humans. Why is it that a young four-week-old dog pup with no previous exposure to humans will readily approach a person, but a wolf pup of the same age and background will run in terror? In this short talk I will discuss my current dissertation work that deals with these questions as well as some of my preliminary findings and their evolutionary implications.      

Kathryn Lord is an alum of Hampshire College, finishing up her Ph.D. at the University of Massachusetts/Amherst in the Department of Organismic and Evolutionary Biology with co-advisors Raymond Coppinger and Melinda Novak. Her research focuses on incorporating development into the study of the evolution of behavior, using dogs and wolves as model species.


February 18: Wendell Kimper (F02) and Emily Elfner (University of Massachusetts).
What can Ned Flanders tell us about linguistic knowledge? Diddly-infixation and the poverty of the stimulus.
What does it mean to know a language? Linguistic grammars are complex systems of rules and constraints that speakers use to create infinite numbers of novel sentences and interpret unfamiliar strings of sounds and words. As children, we acquire our language with no formal instruction, yet we inevitably grow up to be competent speakers with strong intuitions about which utterances belong in our language, and which do not. In this talk, we report on the results of a questionnaire experiment testing speaker intuitions about a novel language-game (diddly-infixation) originating from a speech habit of Ned Flanders, a fictional character the television show The Simpsons. On the show, Ned inserts the nonsense word diddly into regular words of English: for example, WELcome becomes WEL-diddly-ELcome. Diddly is inserted most naturally into words where stress falls on the initial syllable (e.g. WELcome), and part of that syllable is repeated (WEL-diddly-ELcome). Following a hypothesis that the position of diddly and the use of repetition depends on the location of word-stress in the word, we asked our subjects to generalize the process to novel words with initial stress (e.g. CANada), as well as novel words where stress falls on a non-initial syllable (e.g. fanTAStic). We found that speakers agreed not only on how to use diddly in novel words following the initial stress pattern (insertion plus repetition of the syllable, CAN-diddly-ANada), but also agreed on when to deviate from this pattern: in words where a non-initial syllable is stressed (e.g. fanTAStic); our subjects preferred forms with insertion but no repetition (fan-diddly-TAStic) over forms with both insertion and repetition (fanTAS-diddly-AStic). We present an analysis of these results showing that speakers use general linguistic constraints present in their grammar to make decisions about novel linguistic tasks.

Emily Elfner and Wendell Kimper are 3rd year Ph.D. students in the linguistics department at the University of Massachusetts/Amherst. Wendell is a Hampshire alum (F02) and Emily did her B.A. and M.A. work at the University of Calgary.

February 25: Chris Perry, assistant professor of media arts and sciences.
"The Incident at Tower 37": A screening and retrospective .
Over the past four years, the upper-level computer animation curriculum at Hampshire has revolved around the production of a single film. Now that "The Incident at Tower 37" is completed, it is time to take a retrospective look at this unconventional approach to undergraduate animation education. In this talk, I will examine the project from a number of different perspectives in order to better understand its successes and failures. The presentation will feature examples of student work, stories from throughout the production, my own empirical analyses, and of course a screening of the film itself.

Chris Perry, assistant professor of media arts and sciences at Hampshire College, holds an M.S. in Media Arts and Sciences from the Media Laboratory at the Massachusetts Institute of Technology and an M.F.A. in Art from the University of Massachusetts/Amherst. "The Incident at Tower 37" is the second animated short that Chris has directed within the context of a collaborative animation curriculum.

March 4: Noah Charney (University of Massachusetts).
Animal Tracking: An Old Tool for Modern Biologists.
Animal tracking can be a powerful tool in the hands of modern biologists. Some researchers have even proposed that our species evolved in response to natural selection for excellent trackers. In this talk, I will present the foundations of animal tracking. I will offer a look at how tracks, trails, scents, scats, and other sign coupled with natural history knowledge are used to monitor the activities of wild animals. I will also present more high-tech solutions to tracking salamanders that I developed, including a system akin to "EZ-Pass" on toll roads.

Noah Charney is a Ph.D. student in the Organismic and Evolutionary Biology program at the University of Massachusetts/Amherst. His dissertation research examines the terrestrial ecology and conservation of pond-breeding amphibians. He has studied and taught about animal tracking for over a decade, is co-authoring an upcoming book on tracking invertebrates, and is certified in track and sign interpretation.

March 11: Jane Couperus, assistant professor of developmental cognitive neuroscience.
Attention! Visual Selective Attention Across Development: An ERP Study.
We think we know what attention is; we think we know it develops; but what do we really know? We actually know very little about attention, particularly when it comes to children. In this Wednesday’s talk I will be presenting data collected over the summer of 2008 examining attention in children. Previous research suggests that our ability to attend selectively to different aspects of the environment improves across development (Columbo, 2001, Plude, Enns, & Brodeur, 1994, Ridderinkhoff & van der Stelt, 2000, Rueda et al., 2004 ). Less is known, however, about the mechanisms of visual selection, particularly across development. Recent adult research has focused of the locus of visual selection, examining the factors that influence when in processing (early or late) selection occurs (e.g. Lavie, 1995; Lavie & Fox 2000). In particular, perceptual load (i.e., the total available task-relevant information in the visual environment) has been shown to influence the locus of selective attention (e.g. Lavie and Cox, 1997, Lavie & Fox, 2000). In adults, selection appears to occur late in visual processing when perceptual load is low, whereas selection appears to occur early in visual processing when perceptual load is high. Moreover, behavioral studies have suggested the same is true for children (Huang-Pollack et al., 2002). In this presentation, I will present electrophysiological data from studies conducted with children and adolescents (ages 7 to 17) that examine the neurological underpinnings of visual selective attention and the role of perceptual load across development. 

March 25: Cynthia Gill, assistant professor of physiology
Pair-bond formation and stress reactivity in the prairie vole, a model for childhood organization of neuroendocrine and behavioral systems.
Neonatal stress affects physiological and behavioral development. Early separation from parents results in changed neural, endocrine, and behavioral patterns in humans and other animals. In particular, the function of the HPA axis stress-reactive endocrine system is modified by early events, and may be causally related to social behaviors. The prairie vole offers a unique model for examining these changes. Prairie voles form socially monogamous pair-bonds as adults, exhibit bi-parental care of young, and have well-characterized neuroendocrine mechanisms for affiliative social behaviors. Early social isolation evokes stronger plasma corticosterone responses in the pups of the highly social prairie vole than montane voles (Shapiro and Insel, 1990). As adults, their social environment can modulate corticosterone levels and, more excitingly, corticosterone can modulate pair-bond formation (DeVries et al., 1995, Carter, 1998, Ruscio et al., 2007).

We examined the effects of early social isolation on adult pair-bond formation and stress reactivity. We hypothesized that neonatal stress, like stressors in adulthood, disrupts pair-bond formation and adult corticosterone responsivity. During postnatal day (PND) 1-10, pups were removed from their parents for 0 (C), 15 (SS), or 360 (LS) min and housed either individually (PI) or with siblings (PS). Unhandled controls experienced daily lid opening (CU). After PND150, half the animals in each litter were paired with an opposite-sex vole for 24 hours and tested for their preference for their partner over a stranger or neutral area. Stress reactivity was measured in all animals at 0, 30, or 60 min after exposure to a forced swim, with an enzyme assay for corticosterone.

We found that, overall, females and males significantly increased time spent in the partner cage each hour across a 3-hour preference test. Animals exposed to a short separation coupled with housing in isolation as pups, however, showed increased time spent in direct side-by-side contact with their partners as adults. Animals exposed to different separation conditions varied in their choice between the partner or neutral cage, but not the stranger cage. The timing of increased partner interaction also varied across groups. SS/PI females took one hour longer to prefer the partner to the neutral cage. LS/PS males increased partner cage time each hour, though C males did not increase partner time until the third hour. Females spent more time with their partners than males and had higher and more reactive corticosterone levels than males. Long-separated animals had higher baseline corticosterone levels and LS/PI males had a sustained corticosterone response after stress exposure. In agreement that prairie voles are highly sensitive to social variables, preliminary data suggest that adult corticosterone levels correlate with degree of partner bonding more than with exposure to a stressor. Our data further suggest that long neonatal separation disrupts stress reactivity but short separation facilitates pair-bond formation.

Cynthia Gill is assistant professor of physiology at Hampshire College and a member of the Center for Neuroendocrine Studies at the University of Massachusetts/Amherst. She received her Ph.D. from the University of Virginia and did postdoctoral research at the University of Texas. Her research focuses on the developmental regulation of behaviors, specifically through organization of brain and endocrine systems.

April 1: Mara Breen (University of Massachusetts)
How does language learning affect language perception?
Research in language development has demonstrated that, by the age of 10 months, babies have lost the ability to perceive phonemes that are not contrastive in their native language (e.g. Japanese 10-month-olds fail to discriminate /r/ and /l/). Moreover, event-related potentials reveal that adult speakers fail to perceive differences between these phonemes. These data suggest that language experience results in a fundamental remapping of perceptual space, such that non-native phonemes are no longer perceived even at the earliest measurable point. Similarly, by the age of 10 months, babies have learned their native language’ phonotactics, which are language-specific rules that govern how phonemes may be combined (e.g. /dla/ and /tla/ violate the phonotactics of English, and are categorized by native English speakers are /gla/ and /kla/, respectively). There is debate, however, as to whether the inability to differentiate syllables which violate phonotactics reflect a remapping of perceptual space in the same way that it argued for individual phonemes. Prior work by Dehaene-Lambertz, et al. (2000) demonstrated that the ERP waveforms for illegal syllable strings do not differ from those of legal strings at even the earliest measurable point, and therefore argue that language-specific phonotactics have early effects on perception in much the same way that phoneme inventories do. In contrast with this prior work, we demonstrate, in an ERP priming study, early waveform differences between syllables that violate English phonotactics (e.g. /dla/) and those that don’t (e.g. /gla/), despite the fact that these syllables are not differentiated in participants' behavior. We argue that these data provide evidence for a model of perception in which phoneme clusters are faithfully perceived at the earliest stages of perception, but then later categorized according to language-specific constraints (i.e. phonotactics).

Mara Breen graduated from Hampshire College in 2002. Her Division III project was an ERP investigation of the role of phonology in visual word recognition. From Hampshire, she went on to graduate school at MIT, where she studied sentence processing with Ted Gisbon, specializing in prosody. Mara earned her Ph.D. in Cognitive Science from MIT in 2007, and has returned to the Valley as a post-doctoral fellow in Cognitive Psychology at the University of Massachusetts, where she works with Chuck Clifton, Adrian Staub, and Lisa Sanders. Her current work explores multiple aspects of speech production and recognition. Research topics include: (1) acoustic cues to information structure, (2) the role of attention in processing semantic focus, (3) the role of prosody in silent reading, and (4) the relationship between prosody and syntax.


April 8: Jonathan Westphal, adjunct professor of philosophy
The Application of Vector Theory to Syllogistic Logic
Syllogistic arguments can be represented with vectors. If individual premises of a syllogism are diagramed as vectors, then the conclusion of a (valid) syllogism will be the vector sum of the two premises. The result is Leibniz’s Scheme VIII Square of Opposition. The application of the vector representation to Rescher and Gallagher's extension of Venn Diagram techniques to plurative propositions is shown.

Jonathan Westphal studied philosophy at Harvard, Sussex, and London Universities. He has been Professor of Philosophy at Idaho State University for too many years, and he is also a Permanent Member of the Senior Common Room at University College, Oxford. He is a the author of a groundbreaking work on Wittgenstein and colour (Blackwell, 1991, 2nd ed.) as well as eighty or so other publications.

April 15: James Miller, professor of communications
Against Rationality In Journalism and Politics
Serious newspapers are going out of business. Young people pay little attention to the news. Citizens are mostly ignorant about how government works. Few people get directly involved in politics. And much journalism seems to be some version of tabloid, focusing on celebrities (Brangelina as diplomat), the personal (Michelle’s upper arms), the spectacular (Octomom), the scary (any weather report); or fakes (Jayson Blair, Jon Stewart); or blurring boundaries between the real and the pretend (docudramas and Judge Judy).

A common view of all this is to say we’ve collectively gone into a talespin (sic), from the “greatest generation” to the “dumbest generation,” and isn’t it a shame.

This talk considers an alternative interpretation. Changes in news and news consumption correspond to the rise of a cultural citizenship. In post-journalisms and cultural citizenship, the nonrational/emotional/aesthetic triumphs over modernist notions of fact. Politics is at best a secondary experience, whose forms are like those of the rest of pop culture. This is an untidy theory, but promising (and fun).


April 22: Melissa Burch, assistant professor of cognitive development
The Story of Development in Book Reading
Book reading is an activity that many children and parents share. But what can be learned during these interactions? In this talk I will present an overview of some of my research on book reading and discuss how parental support is associated with children’s language and literacy skills. I will also explore how mothers’ conversations during shared book reading change as children’s language abilities increase with age. Finally, I will describe the framework of a collaborative project with the Eric Carle Museum in which teachers have received training in the Whole Book Approach that encourages them to facilitate conversations about the art in picture books. Together, this research allows us to consider book reading as a foundation for the development of language, literacy, and narrative skills,

Melissa Burch is assistant professor of cognitive development. Her research focuses on memory, book reading, and parent-child interaction. She is particularly interested in how adults support children’s emerging memory, language, and literacy skills.