September

Businesss Meeting

Back to Top

J. Leo van Hemmen, Physik Departments, Technische Universität München

Neurobiology of Snake Infrared Vision
How It Can Be Precise Despite Bad Optics

Host: Susan Udin

Back to Top

Jürgen Bohnemeyer, Department of Linguistics, University at Buffalo

How to Hammer a Shirt Apart
(and Talk About It)

We examine the treatment of atypical instrument-theme configurations in events of caused separation in material integrity – i.e., “cutting” and “breaking” (C&B) events - across languages. We focus on four languages which offer an alternative between monomorphemic verb roots and various kinds of complex predicates in the C&B domain: serial verb constructions in Lao and Sranan, compound verb stems in Yucatec Maya, and prefix verbs, particle verbs, resultative constructions, and light verb constructions in German. We test the hypothesis that in accordance with Grice’s Manner maxims, across languages, complex predicates are preferred over simplex predicates for reference to atypical instrument-theme configurations. We draw on descriptions of the “Cut and Break Clips” from five speakers per language. We treat the typicality of a scene as the degree to which it matches the prototype of any linguistic description, and rely primarily on inter-speaker variation as a measure of (a-)typicality in this sense. In line with our hypothesis, we found that the higher the amount of inter-speaker variation a scene elicited in a given language, the more likely the speakers of that language were to prefer complex over simplex predicates. However, though highly significant for the other three languages, this correlation is not significant for German. In German, simplex verbs play no more than a marginal role in the C&B domain; we argue that this upsets the division of labor between simplex and complex expressions underlying the relation between Gricean stereotype and manner implicatures.

Host: Bill Rapaport

Back to Top

Mathias Brochhausen, Department of Philosophy, Unversity at Buffalo & Institute for Formal Ontology and Medical Information Science, Universität des Saarlandes

Building a Cancer Ontology
Bringing Together Clinical Practice and Ontological Theory

The talk will focus on the central question of how practice and ontological theory have to interact in order to provide a coherent and realist account of a complex domain of reality in a way which will allow diverse communities to share and manage data.

We will describe a project to develop an ontology to manage biomedical data concerning cancer research and management throughout the European Union. This provides an excellent example of the techniques used in ontology construction, and also of some of the problems faced in using ontologies to support cross-linguistic integration of data. How do we categorise a domain containing aspects as diverse as molecular biology, patient management, and medical ethics without losing information and yet in such a way as provide a system that can be understood and used by practice-oriented professionals?

Host: Bill Rapaport

Back to Top

Doug Roland, Department of Linguistics, University at Buffalo

Discourse Context and English Object Relative Clauses

While object relative clauses (1) are generally more difficult to process than subject relative clauses (2), a variety of factors such as the pronominal status and givenness of the embedded NP have been shown to reduce or eliminate this difficulty. We argue that this reduction in difficulty is not the result of the pronominal status of the embedded NP per se, but is the result of changes in the relationship between the object relatives and the larger discourse context in which they appear.

In normal language use, object relatives tend to be used to ground the modified NP in the discourse context, while subject relatives are more likely to be used for other purposes. Thus, (2) would be likely to occur whether or not the banker had been mentioned already, while (1) would be likely to occur only if the banker had been the discourse topic. If discourse factors rather than pronominal status or frequency factors were responsible for previous results, the difficulty of object relative clauses with full NPs should be reduced when they are embedded in an appropriate discourse context.

A participant-paced moving window paradigm was used to examine the influence of discourse context on the processing of relative clauses. Preceding context sentences either provided a discourse context where the embedded NP of the relative clause was the topic (e.g., 3), or was related to the target sentence, but did not topicalize the embedded NP (e.g., 4). We found that when an appropriate discourse context was provided, object relatives with full NPs were read as quickly as subject relatives in the regions containing the embedded NP and relative clause verb. However, we also found that, regardless of discourse context, the main clause verb was read more slowly for object relatives than subject relatives.

While providing an appropriate discourse context did not completely eliminate processing difficulties associated with object relatives, given the reading time results at the main verb, it did eliminate previously observed processing differences between subject and object relatives in the relative clause itself. Thus, our results suggest that the pronominal object relative effect may be due, in part, to the better fit between the typical discourse use of object relatives and experimental contexts in which they appeared.

(1) The lady that the banker visited enjoyed the dinner very much.
(2) The lady that visited the banker enjoyed the dinner very much.
(3) The banker was very friendly. The lady that .
(4) There was a dinner party on Saturday night. The lady that .

Host: Bill Rapaport

Back to Top

David Temperley, Eastman School of Music, University of Rochester

Communicative Pressure and Music Cognition

Communicative pressure refers to the pressure on communicative systems to evolve in ways that facilitate the conveyance of crucial information. I begin by presenting some well-documented examples of communicative pressure in language (syntax and phonology). I then examine the relevance of this idea to music. First I explore what I call the syncopation-rubato trade-off: The fact that musical styles with a high degree of rubato (expressive tempo fluctuation) tend to have a low degree of syncopation (misalignment between accented notes and strong beats), and vice versa. I present some recent experimental work exploring the role of the syncopation-rubato trade-off in spontaneous music performance. I then consider another general phenomenon that seems to reflect communicative pressure: Uniform Information Density (UID). The UID principle, first proposed with regard to language Levy & Jaeger 2006), states that producers of communication tend to maintain a relatively even flow of information so as to facilitate processing. I explore several applications of this idea to music, including the correlation between melodic interval size and note length and the effect of harmony on expressive performance. Finally, I consider how the syncopation-rubato trade-off might be explained from the point of view of Uniform Information Density. This talk is geared to a general cognitive-science audience and does not require extensive knowledge of music.

Host: Bill Rapaport

Back to Top

Werner Ceusters, Department of Psychiatry, University at Buffalo

Referent Tracking

Research topics and applications

Referent Tracking (RT) is a paradigm introduced in 2005 intended to provide a means of ensuring unambiguous reference to the particulars in reality that are mentioned in statements. Central to this paradigm is the use of globally unique singular identifiers - called IUIs (for Unique Instance Identifiers) - that stand proxy for the entities in reality to which they refer. For an identifier (ID) to be a IUI, it must refer to one and only one particular, and this tight connection between the particular and its IUI must be asserted by an author in an RT system (RTS). One purpose of the RTS is to give agents who wish to make statements about entities in reality a means to retrieve IUIs for particulars to which identifiers have already been assigned, and to create IUIs in other cases. Another purpose is to provide an efficient way to store data about particulars in terms of their relations to other particulars and to the universals which they instantiate. The applicability of RT has since its inception been assessed in areas such as electronic healthcare record keeping, digital rights management, automated decision support, corporate memories in enterprises, intelligence and outcome assessment. In a recently started project, it is used to bridge the gap between dimensional versus categorical approaches to psychiatric diagnoses.

Of course, RT presents some challenges of its own. One specific problem is how to represent phenomena commonly expressed by statements such as: "no history of diabetes", "hypertension ruled out", "absence of metastases in the lung", and "abortion was prevented". Such statements seem at first sight to present a problem for RT, since there are here no entities on the side of the patient to which unique identifiers can be assigned. Another challenge is keeping track of the different kinds of changes, reflecting for example: (1) changes in the underlying reality, either in a specific patient's condition or the world in general; (2) changes in our understanding; (3) reassessments of what is considered to be relevant for inclusion in a referent tracking database, or (4) encoding mistakes introduced during data entry.

The presentation aims to give an overview of the state of the art in RT. One specific goal is to find amongst the audience researchers that are either interested to contribute to the research, or to apply referent tracking in their research such that we can set up collaborative grant proposals.

Host:Bill Rapaport

Back to Top

James R. Sawusch, Department of Psychology, University at Buffalo

Signal Variability and Perceptual Constancy in Speech
How Listeners Accommodate Variation in Speaking Rate

Human speech is a variable signal because of differences between talkers (dialect, vocal-tract length, habits) and differences within a talker (variable speaking rate, influences of coarticulation). In spite of signal variability, listeners exhibit "phonetic constancy" in recovering the sequence of sounds (phonemes) and words intended by the talker.

The focus here will be on the perceptual mechanisms that listeners use to deal with variability in the signal produced by variation in a talker's speaking rate. A series of studies will be described that examine what type of information talkers embed in the speech signal and the nature of the computations in perception that enable listeners to normalize for variation in speaking rate.

A second series of studies will further examine listeners' normalization in a context of two voices. These studies focus on if/how a listener separates sounds (voices) into different sources, and how the processes that influence sound source separation interact with the listener's adjustment to the speaking rate of the talkers. The behavioral evidence reveal that the perceptual adjustment for speaking rate is an obligatory, autonomous process in perception and is readily influenced by information by voices other than the target talker.

Host: Bill Rapaport

Back to Top

Liz Stillwaggon, Philosophy Department, Univ. of South Carolina & Center for Inquiry, Amherst, NY

Are We Practicing What We Preach?
Methodological Continuity in Cognitive Science

Philosophers are well aware that our beliefs are often out of sync with how we arrive at them. In this talk, I explore one example of how this philosophical problem manifests in cognitive science. A little over a decade ago, the philosopher of mind Peter Godfrey-Smith introduced the related notions of weak continuity (WC) and methodological continuity (MC), i.e., that cognition is an activity of living systems only, and that cognition ought to be investigated only in the context of living systems, respectively. Although many of us might be willing to grant WC, we would be wary of advocating MC because doing so constitutes a fundamental rejection of the project of artificial intelligence. I examine some investigations into the nature of cognition from analytic philosophy, (soft) artificial life, and robotics to determine to what extent they respect (or breach) the principle of MC, and conclude by proposing a methodological guideline that is intended to do justice to both WC and MC

Host: Bill Rapaport

Back to Top

James Collins & Jaekyung Lee, Graduate School of Education, University at Buffalo

Using Writing to Improve Reading Comprehension

This colloquium will discuss findings from the Writing Intensive Reading Comprehension (WIRC) study currently underway in UB's Graduate School of Education. In the WIRC research more than 2000 fourth and fifth grade students in ten low-performing Buffalo schools have taken part in two year-long experiments to test the efficacy of using focused, assisted writing practice to improve reading comprehension. Students enrolled in the WIRC classrooms write daily about the reading they are doing in language arts, and they receive assistance from teachers, peers and thinksheets to help them write about literary selections they are reading. As the name suggests, a thinksheet is a guide to help students think carefully and write about the reading they are doing. The presenters discuss the results of the large experiments which generally show that students in low-performing schools can significantly improve their reading comprehension and writing abilities through sustained, assisted practice with using writing to make sense of their reading. The presenters also discuss formative and qualitative studies designed to help the WIRC researchers understand factors which may facilitate or constrain successful applications of interventions using writing to improve reading comprehension. Findings support further development of sociocognitive theories of literacy development and collaborative means of literacy instruction and assessment.

Host: Bill Rapaport

Back to Top

David Mark, Department of Geography, Unversity at Buffalo

Ethnophysiography
Cross-cultural and Cross-linguistic Variation in Conceptualization of the Landscape and its Elements

Ethnophysiography studies how people conceptualize the natural landscape, especially landforms and water bodies. Are the concepts underlying terms for landscape features and places more or less the same across the languages, or are there significant differences across languages and cultures? Ethnophysiography focuses on kinds of things in the landscape, and aims to document in detail the things in the world (extensions) that are referred to by each term. Ethnophysiography relies heavily on ethnographic methods for obtaining information through interviews, description, and community participation. Because landscape elements seldom fall into anything like 'natural kinds', there is more room for cross-cultural variation than in the cases of plants and animals. Thus the landscape and its elements provide an interesting venue for studies of categorization in general.

For the last several years, my colleagues and I have been conducting ethnophysiography research with the Yindjibarndi people of northwestern Australia and the Dine (Navajo) of New Mexico and Arizona, both of whom live in arid or semiarid landscapes. This presentation will elaborate on ethnophysiography in general, and then present some findings for these two case studies and related work. Almost none of the lexicalized landscape terms in these languages have exact semantic equivalents in the other languages. Yet in most cases the conceptual chunks that are lexicalized in each language seem coherent and reasonable concepts to have terms for. Implications for geographic ontology also will be mentioned.

Host: Bill Rapaport

Back to Top

Adele E. Goldberg, Program in Linguistics, Princeton University

Learning the general from the specific

The constructionist approach to grammar allows both broad generalizations and more limited patterns to be analyzed and accounted for fully. In particular, constructionist approaches are generally usage-based: facts about the actual use of linguistic expressions such as frequencies and individual patterns that are fully compositional are recorded alongside more traditional linguistic generalizations (Langacker 1988). Instances are represented at some level of abstraction due to selective encoding, and generalizations over instances are made as well.

In an impressive interdisciplinary convergence, a similar position has been developed within the field of categorization. Early accounts of categories adopted general abstract summary representations; a subsequent wave of "exemplar based" models of categorization held sway in the field for a period following. Most recently, categorization researchers have argued for an approach that combines exemplar-based knowledge with generalizations over that knowledge. Insights gained from research in general categorization can shed light on how learners go from the specific to the more general. My presentation will focus on several such factors that promote and constrain generalizations: a) skewed input, b) degree of coverage and c) statistical preemption.

Host: Bill Rapaport

Back to Top

December

Michael McGlone, Department of Philosophy, Unversity at Buffalo

Conscious Experience and the Knowledge Argument

Physicalism, as I will understand it, is the view that all correct information is physical information---information of the sort studied by the physical sciences. In this talk I will be concerned with what philosophers call "the knowledge argument" against physicalism. According to this argument, there is correct information regarding the nature of conscious experience that is not itself information of the sort studied by the physical sciences. Although the key idea behind this argument has been with us for a long time, I will focus on a fairly recent formulation due to Frank Jackson (1982). Having considered this form of the argument, I will turn to a recent response, that offered by Jackson himself (2004), and argue that this response is unsatisfactory. Finally, I will attempt to isolate what I take to be the deepest and most interesting issues raised by the argument.

Host: Bill Rapaport

Back to Top