Informational Masking

The energetic vs. informational distinction

According to a loose definition, informational masking (IM) is any form of perceptual masking or interference that cannot be construed as energetic masking (EM). Defined as such, IM is too vast a concept to be of much practical use – just like timbre. This does not mean that the distinction between EM and IM is not important; it certainly provides a good starting point. But if we want to make some progress, we must go beyond it. Oh and, BTW, for those of you who think that the detection of a pure tone in noise involves only EM, read: Lutfi (1990).

The peripheral vs. central distinction

This distinction comes just above the previous one on the scientific ‘utility’ scale. Knowing that IM cannot be explained mainly or solely in terms of peripheral masking is somewhat useful. At the very least, it tells physiologists interested in the neural bases of this phenomenon that they shouldn’t just look in the auditory nerve. Unfortunately, contrary to the once predominant view among auditory scientists, the central auditory system is a lot more than just an appendix to the cochlea; so, knowing that IM is central still leaves us with a world of possible mechanisms to explore. In addition, it’s worth noting that, even though IM undoubtedly involves central phenomena, peripheral functional properties such as frequency selectivity can nonetheless play an important role in determining it: in particular, the amount of IM in experiments where the effect appears to be due chiefly to listener’s uncertainty can be predicted (quantitatively and rather successfully) by the ensemble variance of peripheral auditory filter outputs evoked by randomly-varying maskers (see Lutfi, 1993). From that point of view, the peripheral vs. central distinction starts to fade.

The low- vs. high-uncertainty distinction

This marks the great divide between the “purists”, who think that true IM only arises in conditions where the listener is highly uncertain about what the next stimulus is going to be like, and the “generalists” who count as IM perceptual interference or masking that cannot be EM, yet is observed under conditions where uncertainty is minimal. For instance, the latter might categorize MDI or other forms of across-channel interference as IM. However, we would like to suggest that the notion of “uncertainty” contributes to blur things rather than to clarify them. Until uncertainty can be defined precisely and measured in another way than through IM (which otherwise leads to a circular argument), the notion should be abandoned in favor of physical and quantifiable notions like the degree of stimulus variability or the number of possible stimuli that can occur during the course of an experiment.

The synthetic vs. analytic distinction, selective attention, and perceptual organization

For some, IM is a failure of selective attention to the target. While this may be true, it does not help much, unless we can more easily determine why selective attention to the target fails than we can determine why IM occurs. Unfortunately, at this point, the factors that determine a listener’s selective attention seem to be at least as obscure as those that determine IM. Let’s note also that there are different ways in which selective attention to the target can fail. The listener’s attention can be distracted away from the target by one or more of the masker components (i.e., the listener is still attending selectively but to the wrong thing). Alternatively, the listener may be unable to perceptually separate the target from the masker components, and IM may then result from the resulting global (synthetic) percept being dominated by the more energetic and numerous masker components. Looking for ways of experimentally measuring the respective contributions (if any) of these two types of mechanisms (i.e., distraction and perceptual grouping) seems like a necessary avenue for research on IM. One interesting implication of the notion that IM depends upon perceptual organization (i.e., little or no IM if the target can be perceptually segregated from the masker), is that knowing what factors promote segregation, we would also know what factors should reduce IM. Let’s note in this respect that perceptual segregation depends not just on the properties of the stimulus, but also on the listener’s experience (internal templates or memory traces). The distinction between analytic and synthetic listeners appears to imply that listeners differ greatly in their ability to form or utilize templates. The question, then, is why should they? Prior experience comes to mind, and while some “synthetic” listeners appear completely impermeable to protracted psychoacoustical training (Neff and Dethlefs, 1995), musical training appears to favor analytic-listening abilities (Oxenham et al., 2003), suggesting that at least some forms of training can influence IM.

Exploring the perceptual and neural mechanisms of IM: start simple

We’d like to conclude this statement by arguing for a “start-simple” philosophy, which we are currently putting in practice as we’re trying to better understand (and build models of) the perceptual and neural mechanisms of another vast and complex aspect of auditory perception: auditory scene analysis. Rather than trying to embrace the phenomenon in all its variety and complexity, we start by selecting a simple experimental paradigm, which captures some basic aspect of the perceptual phenomenon under study. As it turns out, one of the paradigms that are working with comes from the IM literature, and involves the detection of repeating target tones among randomly varying, multi-component masker tones (Kidd et al., 1994; 2003). This paradigm meets all the above requirements: it minimizes the role of peripheral EM (thanks to the use of a protected spectral region around the target), produces high uncertainty (with spectro-temporally varying maskers), and allows us to explore the respective contributions of selective attention, and perceptual organization. Even with such a simple paradigm, there are numerous stimulus parameters to control, and many factors that can influence perception, including for instance the average frequency separation between the target and masker components, their repetition rate, temporal regularity,… Yet, with careful stimulus design, the respective contributions of different factors can be determined. Preliminary results suggest that performance is determined by only a few factors, such as the average frequency separation between the target and the maskers and their relative repetition rates. Other factors, such as temporal regularity, appear to play only a marginal role in the listener’s ability to hear-out the repeating target tones among the random multi-tone maskers. Thus, the phenomenon may involve less complex and varied mechanisms than it may have seemed at first. These psychophysical results are then used to guide the search for neurophysiological correlates of the perceptual phenomenon in the primary auditory cortex, and inspire the design of models involving relatively simple neural mechanisms such as frequency selectivity, lateral and forward inhibition, and adaptation. The relative success of this simple-minded approach to the exploration of the neural basis of auditory stream segregation in recent years (e.g., Bee and Klump, 2004; Fishman et al., 2004; Micheyl et al., 2005) leads us to belief that a similar strategy may also benefit the understanding of IM.

References

Bee MA, Klump GM. (2004) Primitive auditory stream segregation: a neurophysiological study in the songbird forebrain. J Neurophysiol. 92, 1088-1104.

Fishman YI, Arezzo JC, Steinschneider M. (2004) Auditory stream segregation in monkey auditory cortex: effects of frequency separation, presentation rate, and tone duration. J Acoust Soc Am. 116, 1656-1670.

Kidd G Jr, Mason CR, Deliwala PS, Woods WS, Colburn HS (1994) Reducing informational masking by sound segregation. J Acoust Soc Am 95, 3475-3480.

Kidd G Jr, Mason CR, Richards VM. (2003) Multiple bursts, multiple looks, and stream coherence in the release from informational masking. J Acoust Soc Am. 114, 2835-2845.

Lutfi RA (1990) How much masking is informational masking? J Acoust Soc Am. 88, 2607-2710.

Lutfi RA (1993) A model of auditory pattern analysis based on component-relative-entropy. J Acoust Soc Am. 94, 748-758.

Micheyl C, Tian B, Carlyon RP, Rauschecker JP. (2005) Perceptual organization of tone sequences in the auditory cortex of awake macaques. Neuron 48, 139-148.

Neff DL, Dethlefs TM (1995) Individual differences in simultaneous masking with random-frequency, multicomponent maskers. J Acoust Soc Am. 98, 125-134.

Oxenham AJ, Fligor BJ, Mason CR, Kidd GJr (2003) Informational masking and musical training. J Acoust Soc Am. 2003 114, 1543-1549.