In a dichotic listening task where participants listen to different (frequency-wise) musical tones through headphones, is cross talk to the contralateral cochlea of each ear through bone conduction a source of noise to be taken into consideration, or is it insignificant? If it is a source of noise, are there any ways to eliminate physically or computationally any such effect?


Crosstalk needs to be taken into consideration, but with reasonable controls it can be made insignificant. Interaural attenuation can vary between less than 40 dB to over 60 dB depending on frequency, the type of headphone, and how well it is coupled to the head. If you have 40 dB of interaural attenuation and present the sounds at less than 40 dB SL (sensation level) the cross talk will be inaudible. The further below threshold, the better off you are.

If you need to present the sounds at a level greater than the amount of interaural attenuation, crosstalk will still not matter if the other sound provides sufficient masking or you introduce sufficient masking noise. For dichotic speech tasks, the speech masks the crosstalk. For musical tones, you might need to introduce a masking noise depending on the frequency content of the tones.

It may be easier to simply spatially separate the stimuli by either presenting them over loudspeakers or using head-related transfer functions (HRTFs) and presenting them over headphones. In the original work of Broadbent (1954) he used a number of different configurations while Cherry (1953) only used headphones and presented each stimulus monaurally.

  • $\begingroup$ Thanks for the answer. My thoughts were on masking too, but given the fact that I want to measure valence with regard to simultaneous musical intervals I am a bit worried if a masking noise would effect participants' judgements. Also if a tone is inaudible due to masking, is ti safe to assume that its spectral components will not interact with those of the other tone in a way that would give rise to beating? $\endgroup$ Feb 1 '15 at 11:12
  • $\begingroup$ @LarryM. there will not be any beating if you add enough masking noise, but this of course could screw up your experiment. Why not just spatially separate the sources (see edit)? $\endgroup$
    – StrongBad
    Feb 1 '15 at 13:13
  • $\begingroup$ The first link is probably behind a paywall and the second says "page not found". I guess the second link should lead here. It may not be clear in my question but the spatial separation (if I understand correctly that is left and right panning for the different tones) is what I wanted to do from the start regarding the presentation of the stimuli. But, without using a masking noise or lowering the SPL for both tones, the problem of crosstalk will probably remain. Nevertheless, your answer was much helpful. $\endgroup$ Feb 1 '15 at 22:03
  • $\begingroup$ @LarryM. I fixed the links. Cross talk is not a particularly bad thing, unless you want to say the stimulus is only activating a particular cochlea. Panning is in essence just changing the amount of cross talk. $\endgroup$
    – StrongBad
    Feb 2 '15 at 12:05
  • $\begingroup$ Exactly, I don't want, for example, a tone that is presented through headphones to one ear to stimulate the cochlea of the other ear by means of bone conduction. I am sorry if I was unclear. $\endgroup$ Feb 2 '15 at 17:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.