Abstract :

General recognition theory was developed from multi-dimensional signal detection theory by Ashby & Townsend to provide a rigorous theory and related methodology that could assess decisional and perceptual separability, and perceptual independence within a uniform framework.
It is being used in many fields to study relationships among psychological dimensions and features. In this talk, I introduce the basic foundations in a logical but relatively non-technical fashion and indicate how one employs the fundamental strategies.

Résumé :

Dans le monde réel, nos mouvements sont des événements dynamiques qui impliquent de la matière et de l'énergie. Dans les mondes électroniques synthétisés par les ordinateurs, en revanche, nos mouvements échappent dans une large mesure aux contraintes de la dynamique. Dans un atlas géographique, par exemple, il n'y a pas de limite supérieure à la vitesse de notre locomotion virtuelle (le mouvement d'une vue dans un atlas virtuel peut fort bien excéder la vitesse de la lumière). Nous discuterons du statut de la dimension d'échelle- par opposition à une dimension spatiale- et nous rapporterons les résultats d'une expérience sur la loi de Fitts dans laquelle les sujets, équipés d'un zoom, ont dû pointer sur des cibles extrêmement difficiles en combinant des déplacements spatiaux (pans) et de changements d'échelle (zooms). Les données montrent que dès lors que l'échelle de l'interaction devient une variable libre (une possibilité spécifique aux mondes d'information, qui sont des mondes purement cinématiques, sans masse), il n'y a plus, en principe, de limite supérieure au pouvoir différenciateur de nos mouvements : D désignant la distance à couvrir pour atteindre une cible de diamètre W, n'importe quel rapport D/W devient maîtrisable (nous avons exploré jusqu'à 109 dans notre étude, bien au delà du rapport maximal D/W = 500 de la littérature). En second lieu, nous montrons que la loi de Fitts, qui lie le temps de mouvement (TM) au rapport D/W, se généralise à n'importe quelle valeur de ce rapport, tout en se simplifiant : nos données suggèrent la loi TM = k.log2 (D/W).

Abstract :

The understanding of human auditory cortical function can be approached from different perspectives - psychological functions embedded in the frameworks of perception and cognition such as comprehending spoken speech, psychophysical discrimination thresholds and the physiology of the cortical areas, which may be studied in analogy to the animal literature. Brain-behavior relationships for complex perceptual and cognitive auditory functions have traditionally been studied in the field of neuropsychology, in which deficits have been delineated in neurological patients suffering from circumscribed brain lesions.
Typical patients suffering from disorders such as auditory agnosia or cortical deafness will be described. With modern non-invasive functional imaging techniques, most notably PET, fMRI, and MEG, auditory cortical functional anatomy can now be studied in healthy human subjects in vivo.
Results of such studies will be described, with particular emphasis on the processing of complex non-verbal sounds such as musical tones and environmental sounds. Recent technological developments will be described.

atelier "Magnétoencéphalographie et audition"

le programme
C'est une série de présentations sur la psychoacoustique donnée par des membres du département de Psychologie à l'Université d'Essex en Angleterre.

Abstract :

The ear acts like a spectrum analyzer, mapping sounds of different frequencies to different locations in the cochlea; hair cells at a given location along the basilar membrane (BM) respond maximally to tones of a characteristic frequency. The encoding of sound level depends on the level of BM vibration. The BM in normal hearing is markedly non-linear. It seems that normal hearing depends on BM non-linearity whereas some types of sensorineural hearing loss, e.g. loudness recruitment, are associated with loss of non-linear function. A project is underway at the University of Essex to measure BM non-linearity which will help refine models of hearing function in both normal hearing and hearing-impaired listeners. Techniques for estimating BM compression utilizing forward masking will be described and results illustrating the effects of compression will be given.

Abstract :

Comodulation Masking Release (CMR) is a phenomenon where the detection of a
masked signal is improved by the addition of other sounds, provided that the added sounds have the same emplitude envelope fluctuations over time as the original masker. Typically benefit to detection is observed when the maskers are co-modulated, that is, have the same amplitude fluctuations over time, across frequency. The present work tests hypotheses regarding the role of forward-masking type cues in CMR paradigms using pure tones presented with a sinusoidally amplitude modulated (SAM) pure-tone masker.
Specifically, brief signal tones were presented in the amplitude minima of the envelope of an on-frequency SAM masker. With this method of presentation, the primary form of masking is non-simultaneous, as the signal is masked by preceding (forward masking) and following (backward masking) amplitude envelope peaks. It is possible that the addition of a comodulated flanker may provide a release from non-simultaneous masking by a) cueing the listener as to the ideal time to listen for the signal, or b) suppressing the masker. In a series of experiments, it was found that there was a release from masking with comodulated flankers that is consistent with a cueing or confusion explanation for CMR. There was a greater release from masking with ipsilateral than contralateral flankers, providing some evidence for a role of suppression. Furthermore, when a single flanking tone is used, the pattern of masking release observed with an ipsilateral flanker varies with flanker tone frequency in a manner consistent with suppression effects. When compared to unmodulated control conditions, there is greater release from masking with comodulated than with unmodulated stimuli. This result indicates that there is an additional benefit of modulation per se.

Abstract :

Complex tones are sounds with repetitive waveforms that consist of a number of harmonic frequency components. These tones are often associated with a distinct pitch that usually corresponds to the repetition rate (or fundamental frequency, F0). The first ten harmonics of a complex tone (approximately) can be separated out by the cochlea and are said to be "resolved". Higher harmonics interact in the cochlea to produce a composite waveform that repeats at the repetition rate of the complex. In the past, pattern recognition models assumed that pitch is derived from the patterning of the resolved harmonics, whereas temporal models assumed that pitch is derived from the interacting unresolved harmonics. Recently, however, "single mechanism" accounts of pitch perception have become popular. These models combine the information from the resolved and unresolved harmonics, often by utilising the auto-correlation function (ACF). In this presentation it will be argued that there are differences in the processing of resolved and unresolved harmonics, particularly with respect to the time constant(s) of the pitch mechanism(s), that suggest that a single mechanism account may not be viable.

Conjointement avec le séminaire du Laboratoire de psychologie expérimentale

Elyse Sussman 1 & István Winkler 2 Effects of predictability and unpredictability on the processing of sound changes
1 Department of Otolaryngology, Albert Einstein College of Medicine, NY, USA / 2 Institute for Psychology, Hungarian Academy of Sciences, Budapest, Hungary

Abstract :

Sound change is an important cue for the auditory system, often providing new information. Sensory changes that are predictable to a given brain process, however, carry no new information for that process. We will discuss two experiments that show effects of predictability on the processing of sound changes. In one case, predictability results from stimulus driven characteristics of the sound sequence (bottom up) and in the other the predictability results from knowing in advance when a sound change will occur (top down). The results showed that predictability affected sound changes at different levels of cognitive processing.