January

Businesss Meeting

Back to Top

Randall Dipert, Department of Philosophy and Center for Cognitive Science, University at Buffalo

Game Theory for Philosophers and Cognitive Scientists

Advanced mathematical results in its various branches have not generally had much impact in philosophy.  (The story with cognitive science is much more complicated.) The notable exceptions are geometry, number theory, and the theory of computation (up to 1940).  One simply does not know how most other, sometimes deep, results would affect various philosophical questions, if at all.  One can guess that various results in complexity theory (e.g., NP Completeness), information theory, and combinatorics (especially graph theory) might have some philosophical impact.  A minor exception is the use of game theory in David K. Lewis theory developed in his widely acclaimed book, Convention (1969).

Host: Bill Rapaport

Back to Top

David Hunter, Department of Philosophy, Ryerson University

Dispositions and Psychological Explanation

It seems clear that belief makes a difference to behavior; that agents differ in what they believe only if they differ in their behavioral dispositions. But certain cases of first-person belief raise a serious puzzle for this view. They are cases where agents with the very same beliefs, or at least who agree on all the facts, nonetheless act differently. Some theorists have responded by proposing to individuate beliefs more finely than by agreement on the facts, concluding that agents who agree on everything might nonetheless differ in their beliefs and consequently in their behavioral dispositions. I will argue that this is the wrong lesson to draw from those puzzling cases, and will sketch what I think is a better lesson.

Host: Bill Rapaport

Back to Top

Peter Q. Pfordresher, Department of Psychology, University at Buffalo

Why can't Johnny sing?

Mechanisms for vocal imitation of pitch

Despite the seeming prevalence of "poor pitch" singing, basic questions regarding its cause, prevalence, and symptoms remain unanswered. I will discuss the results of recent research designed to form an objective description of "bad" singing and to test hypotheses regarding its origin. In the experiments I report, participants (typically non-musicians and non-singers) complete a variety of tasks that involve vocal production or perception of musical pitch. Contrary to the implications some recent research, bad singing does not seem to result from "tone deafness" literally speaking: Bad singers are as good as good singers on basic perceptual tasks. In addition, it seems that bad singing cannot be exclusively related to control of laryngeal muscle. Instead, our data suggest that inaccurate singing results from inappropriate mapping of pitch information onto motor gestures, potentially in neural systems reserved for the mental simulation of action.

Host: Bill Rapaport

Back to Top

J. David Smith, Department of Psychology and Center for Cognitive Science, University at Buffalo

Prototypes

and the Emergence of Cognitive Systems for Categorization

I will explore--from a utility perspective--the role of prototypes and exemplars in learning natural-kind or family-resemblance categories. The organization of the psychological spaces containing these categories determines the felicity of categorizing using different representations. Humans and nonhuman animals will receive more variable and less useful signals about category belongingness if, as exemplar theory supposes, they store individual exemplars separately and use these as the| comparative standard for categorization. Humans and animals will receive more stable and useful signals about category belongingness if they average their exemplar experience, abstract the prototype--as prototype theory supposes--and use this as the comparative standard for categorization. Thus, category members and non-members--to be approached and avoided, perhaps--will be more discriminable when referred to a prototype and therefore categorized more accurately and adaptively. There are strong reasons for the cognitive systems that serve categorization to have come to include prototype abstraction among their capabilities.

Host:Bill Rapaport

Back to Top

Eric Dietrich, Department of Philosophy, Binghamton University

The Bishop and Priest

On The Epistemology and Psychology of True Contradictions

True contradictions are increasingly taken seriously by philosophers and logicians. Yet, that contradictions are always false remains deeply intuitive. This paper presents a psychological model explaining how certain contradictions can be seen to be true. The contradictions in question are called "limit contradictions," and have been systematically studied by Graham Priest. The specific limit contradictions we will look at are those associated with Bishop George Berkeley's idealism. The paper's goal is to weaken the intuition that contradictions are always false, while planting an intuition that some contradictions are true. If philosophers could psychologically understand how limit contradictions can be true, beyond just saying that the contradictory conclusions follow validly from true premises, then perhaps that they are true would become more tolerable.

Host: Bill Rapaport

Back to Top

Barbara Landau, Department of Cognitive Science, John Hopkins University

Starting at the End

The importance of goals in spatial language and spatial cognition

A hallmark of human cognition is our capacity to talk about what we see. How is this accomplished? Given that language and spatial representations are likely to have quite different kinds of structures, the challenge is to understand how such apparently different systems of knowledge map onto each other, and how these mappings are learned. In my talk, I will discuss this problem with respect to the language of events, including manner of motion, change of possession, attachment/ detachment, and change of state events. I will focus on evidence from normally developing children and children with Williams syndrome, a rare genetic deficit that gives rise to an unusual cognitive profile of profoundly impaired spatial representations together with spared language. The evidence shows that a fundamental property of event semantics, an asymmetry between source and goal expressions, is a pervasive fact about the linguistic description of events. Ancillary evidence suggests that this asymmetry is also a part of our non-linguistic representations, appearing in non-linguistic tasks among infants, children, and adults. As a whole, the results suggest a homology between spatial language and spatial representation, thereby providing a partial solution to the problem of mapping dissimilar domains onto each other.

Hosts: Gail Mauner, Dept. of Psychology, UB, and Jean Pierre Koenig, Dept. of Linguistics, UB

Back to Top

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

Canceled

Host: Susan Udin

Back to Top

Steve Petersen, Department of Philosophy, Niagara U.

Software Intelligence

These days we are happy to say at least figuratively that some bits of software are intelligent. Chess programs, spam filters, bots on the emerging semantic web, and villains in computer games are all getting "smarter". Though merely metaphorical now, we might wonder if it is possible, someday, for a piece of pure software to be /literally/ intelligent--intelligent in the same important sense normal humans are.

Much hangs on this question. According to a growing consensus, intelligence has to do with adaptability in the face of environmental goals--and this notion has motivated much of the "embodied" approach to artificial intelligence. The possibility of software intelligence complicates this consensus, though, and the accompanying embodied robotics program. If software intelligence is possible, for example, probably it will be significantly easier and cheaper to engineer than "real-world" robots. There are ethical implications to the thesis as well, since presumably anything with genuine intelligence carries at least some moral significance. A simulation involving virtual agents in a natural disaster could someday be just as horrific as engineering a natural disaster in the "real world".

I argue that even on the consensus view of intelligence, such "software intelligence" is indeed possible.

Host: Bill Rapaport

Back to Top

April

Jim Swan, English Department, University at Buffalo

The Long and Short of It
Speech Rhythms, Prosody,
and Cross- Linguistic Effects

The fundamental place of rhythm in speech is well understood. It has been shown, for instance, that people in conversation attune themselves to one another’s rhythm and intonation; that neonates, only a few days old, alert to the sound of their mother’s language and not to others; that language use, the production of meaning, is inescapably rhythmical.

This talk is the beginning of a study of prosody in poetry, specifically of the effect that translating or adapting work in another language has on a writer’s practice in her own language. Here I will be focusing on an important historical moment when English poetry finds its early-modern footing after medieval English and its strongly French-influenced pronunciation falls from use. English writers in the late 16th century grow up learning to read and write Latin poetry and prose. Prosody in classical Latin is markedly different from prosody in English, and when a group of poet- experimenters try their hand at writing poems in English using classical meters, the result is far-reaching and profound. My argument is that, through these experiments in the formal properties of another language, poets discovered the deep prosodic potential of their own language. In a longer version, I would extend the analysis forward to the beginnings of 20th-century modernism, when Ezra Pound experimented with translations of Old (i.e., pre-medieval) English and Homeric Greek poems.

In my analysis of lines and phrases, I will surely reveal my naiveté about matters phonological, and I look forward to comments and help from the colloquium.

Host: Bill Rapaport

Back to Top

Eduardo Mercado III, Department of Psychology and Center for Cognitive Science, University at Buffalo

Learning and Representation in Auditory Perception

Learning to discriminate and identify sounds can change how an organism’s brain responds to those sounds, and how easily differences between similar sounds are recognized. Recent efforts to rehabilitate children with language processing difficulties have attempted to take advantage of this phenomenon by using progressive auditory training programs to improve individuals’ perceptual sensitivity to differences in speech sounds. However, a current lack of understanding about how perception and learning interact makes it difficult to predict: (1) how effective such strategies might be for a particular individual, (2) to what extent developmental factors that constrain the effects of learning on perception can be circumvented, or (3) whether different percepts within or across modalities might be more or less plastic. In this talk, I will discuss recent comparative studies of auditory perceptual learning in rats and humans that shed some light on the mechanisms that constrain how learning experiences impact perceptual abilities. Studies with adult rats show that some auditory distinctions are extremely difficult or impossible for rats to learn to make unless they have prior experience with less challenging distinctions, suggesting that the fidelity or utility of auditory
| representations is experience-dependent. Electrophysiological measures of cortical responses from naïve and experienced rats suggest that auditory training increases the selectivity of cortical neurons for particular acoustic features experienced during training. Behavioral experiments with humans also indicate that different auditory training experiences can differentially impact perceptual sensitivities. Scalp recordings collected in parallel with this behavioral task suggest that behavioral improvements parallel changes in cortical response properties. Interestingly, the largest changes in cortical responses occurred for sounds that were already easily distinguished before training, suggesting that changes in cortical representations continue to occur even when behavior is stable. Additionally, the best neural predictor of an individual’s ability to distinguish two sounds was not the response evoked after repeated presentations of one sound followed by the presentation of the second sound (the ubiquitous oddball task), but instead the response evoked when only one of the sounds was repeatedly presented, again suggesting that representational selectivity constrains performance. Cortical correlates of auditory perceptual learning point to adaptive spatial isolation in cortical networks as a primary mechanism of perceptual differentiation.

Host: Bill Rapaport

Back to Top

Myrna Schwartz, Moss Rehabilitation Research Institute

Beyond double dissociation

The graceful degradation and recovery

of language production in aphasia

Cognitive neuropsychology has traditionally sought information about the modular mind/brain systems that support cognition, using evidence from double dissociations. A complementary question is whether brain systems, modular or not, display the property of graceful degradation, meaning the ability to approximate the desired response in the face of imperfect information or when damaged. This talk has three aims: (a) to explain the graceful degradation hypothesis in relation to connectionist models of language production; (2) to present data from patient studies that supports its validity; and (3) to show that double dissociations and graceful degradation together place strong constraints on models of language processing.

Host: Gail Mauner, Dept. of Psychology, UB, and Jean Pierre Koenig, Dept. of Linguistics, UB

Back to Top

Eva Juarros-Daussà, Department of Romance Languages and Literatures, University at Buffalo

Lack of Recursion in the Lexicon
The Three-Argument Restriction

It is a universal property of argument structure that verbs cannot take more than three arguments –one external and two internal--, as in the English verbs give or put, unless if introduced either by an extra lexical preposition (1a), an applicative or causative morpheme (1b), or the additional head of a serial verb construction (1c) (Dixon and Aikhenvald 2000, Peterson 2007). In any case, the valency of the resulting predicate is restricted to that of a ditransitive verb. I call this the Three-Argument Restriction (TAR, see 1d), which remains one of the mysteries of human language. Since there is no known processing reason not to lexically associate more than three participants to a predicate, I claim that the TAR is syntactic in nature, and is one of a family of architectural constraints that determine and limit possible attainable languages (in this case possible argument structures). In this talk I examine the TAR and its challenges, including a case study involving Romance clitics. I further examine how such generalization could be derived within the framework of lexical syntax put forth by Hale and Keyser 2002. According to this proposal, deriving the TAR crucially involves negating the existence of a recursive function in the domain of argument structure, relegating such functions to the domain of sentences.

1a. Erika gave Alexandra a present *(through/on behalf of) Frances

1b. wa’-khe-yó:nye’        Meri  a:y-e-yo-h          gwa’yea’
    FACT-1sg.NOM.3ACC- make-PUNC Mary OPT-3.SG.NT.NOM-kill-PUNC rabbit

“I made Mary kill the rabbit”

(Baker 1996)

1c. Olú ti       omo   náà  subú

      Olu push child  the   fall

“Olu pushed the child down; Olu caused the child to fall by pushing him”

(Baker 1996)

d. “A single predicate must have at most two internal arguments and one external”

Host: Jean Pierre Koenig

Back to Top