Detection thresholds for brief sounds ^ are they a measure of auditory intensity integration? Katrin Krumbholz *, Lutz Wiegrebe Zoologisches Institut der Ludwig-Maximilians-Universita ët Mu ënchen, Luisenstrasse 14, 80333 Munich, Germany Received 28 July 1997; revised 27 June 1998; accepted 27 June 1998 Abstract The present study measured how the detection thresholds for two very brief tone or noise bursts depend on the delay between them. The thresholds for the tone burst pairs systematically increase for increasing delays up to a few milliseconds and then reach a constant value. The tone burst thresholds thus reflect a very short-term interaction between the brief signals. Strikingly, the time constant of the interaction appears to scale inversely to the frequency of the tone bursts. The thresholds for the noise burst pairs, on the other hand, remain approximately constant down to the shortest measured delay of less than 1 ms. In contrast to the concept of intensity integration, these results are interpreted in terms of a temporal overlap of the auditory filter responses elicited by two brief, successively presented signals. In a second experiment it was shown that the observed short-term interaction between two brief signals is sensitive to their relative phases, corroborating the conjecture that the interaction takes place at a level prior to the mechanical to neural transduction in the inner ear. z 1998 Elsevier Science B.V. All rights reserved. Key words: Intensity integration ; Temporal interference ; Decay time of auditory ¢lter ; Matched ¢lter 1. Introduction The auditory system can integrate information that is distributed over time, a phenomenon referred to as tem- poral integration (e.g. Green et al., 1957; Plomp and Bouman, 1959; Florentine et al., 1988). Swift £uctua- tions of the instantaneous amplitude of sounds, on the other hand, are not perceived with an arbitrarily ¢ne resolution (e.g. Plomp, 1964; Forrest and Green, 1987; Shailer and Moore, 1983). These two contrary aspects of temporal auditory processing ^ temporal integration and temporal resolution ^ are often being described with the concept of a central integration process that integrates over the instantaneous amplitude of the pe- ripherally pre-processed signals (de Boer, 1985). Due to the fact that the detectability of deterministic and spec- trally static signals is roughly proportional to signal energy (Green et al., 1957), the input to the integration process is often considered to be proportional to the instantaneous intensity of the signals. In some models, however, a compressive function of sound pressure is integrated (Penner, 1978; Oxenham and Moore, 1994; Moore et al., 1996). The integrator can be regarded as a linear ¢lter with a ¢xed temporal weighting function. The general shape of the integration weighting function di¡ers in the various models. Well-known examples are the leaky-integrator model which treats past sound events with an exponentially decaying weight (Zwis- locki, 1960, 1969; Plomp and Bouman, 1959), the per- fect intensity integrator which is represented by a rec- tangular weighting function of ¢nite duration (Green et al., 1957; Green, 1960) or the temporal window pro- posed by Moore et al. (1988) having an asymmetrical rounded-exponential (roex) shape (see also Plack and Moore, 1990). In the integration models, the decision mechanism operates on the output of the integrator and usually depends on the nature of the experiment that is to be described (Oxenham and Moore, 1994; Forrest and Green, 1987; de Boer, 1985). 0378-5955 / 98 / $19.00 ß 1998 Elsevier Science B.V. All rights reserved. PII:S0378-5955(98)00134-8 * Corresponding author. Tel.: +44 (1223) 333857; Fax: +44 (1223) 333840; E-mail: katrin.krumbholz@mrc-cbu.cam.ac.uk Hearing Research 124 (1998) 155^169