摘要:AbstractVisual and auditory duration information plays an important role in our perception of external events. However, mismatches frequently occur in the temporal extent of the event's auditory and visual components (e.g., seen and heard elements of striking a cymbal). Moreover, these physical discrepancies will be accompanied by perceptual differences arising from internal neural noise. An efficient system for integrating these signals might resolve these discrepancies by fusing the signals into a multisensory estimate of audiovisual duration. However, this approach would lead to potentially hazardous misperceptions if large audiovisual discrepancies originate from separate external sources. In the current study, we investigated the role of duration discrepancy in regulating the multisensory perception of duration. Observers made visual duration discrimination judgments (centred on 320ms) whilst attempting to ignore a range of concurrently presented auditory ‘distracter’ stimuli (100-640ms in logarithmically spaced steps). In line with earlier reports (Klink, Montijn & van Wezel, 2011), small duration discrepancies caused perceived visual duration to be biased in the direction of the auditory distracter. Small increases in auditory distracter duration (>320ms) induced approximately linear increases in perceived visual duration until ∼0.2 log units of duration discrepancy, beyond which the influence of the auditory distracter declined rapidly. There was no evidence of a similar tuning pattern with an equivalent decreases in auditory distracter duration (<320ms): perceived visual duration showed robust contraction with our shortest distracter. Given that duration discrepancies were matched (in logarithmic terms) either side of our 320ms visual stimulus, our asymmetric tuning pattern cannot be driven by relative durationper se.Alternatively, multisensory integration may be driven by onset/offset asynchronies which were disproportionately greater for longer auditory distracter/visual stimulus combinations. A follow up experiment confirmed this possibility: visual judgments were made around a 640ms baseline accompanied by a range of distracter stimuli which generated equal duration discrepancies either side of 640ms (in log terms) but substantially greater onset/offset asynchronies across the range. In this situation, there was little evidence of robust audiovisual interaction, suggesting that discrepant onsets and/or offsets are a more powerful perceptual segregator than the durations bound within them.