As is perhaps obvious, people really are very different from pieces of coal and blocks of ice – if we try to measure their luminance threshold on Tuesday we will
most likely get a different result to the measurements we take on Thursday. More pressing, however, is that if we estimate threshold at 11.00 on Tuesday, this value is likely to change if we re-test at 11.05. As Haber and Hershenson (1979) argued, this may be because we may have poor control over stimulus factors – our light source may vary in an erratic fashion. More importantly, though, we may have very little control over subjective factors. The participant might be bored/ excited, tired/alert, happy/sad, drunk/sober and so on. The participant’s psychological state can vary and, as experimenters, we probably will have little (possibly no) control over this. It is therefore highly unlikely that we will ever record a step function with a fixed threshold for some perceptual decision task. It is far more likely that the so-called psychophysical function relating accurate reports to a given stimulus variable (such as luminance) will be best approximated by a continuously smooth curve (see Figure 4.6b). If we admit this, however, we then have to decide where to place the threshold.
So one way to proceed is to design an experiment whereby on each trial we present the participant with a spot of light and simply get them to report what they see. Over trials we vary the intensity of the light (how dim or bright the light is) and note the participants’ responses. (Did they see that very dim light just pre- sented or did they not?) Assuming that a value of 0 signifies the light is off and 7 signifies the light is fully
In order to provide evidence for the success of the subliminal speech, participants should have been quicker to respond to the target when the prime and target were related than when there was no relation between the prime and target words. The data showed that when the participants failed to identify the prime (providing at least some evidence for the subliminal nature of the effect), subliminal speech priming was restricted to words rather than non-words, and was apparent only for cases where the prime was a complete repetition of the target. In contrast, when the participants reported hearing the prime (providing evidence for the nature of effects when the prime is supraliminal), then par- ticipants were faster to respond to target words with semantic and morphological primes. With clearly heard primes, non-word complete repetition priming was also in effect.
Despite some differences between sound and vision, Kouider and Dupoux (2005) concluded by
drawing parallels in subliminal priming between the auditory and visual modalities. For instance, both modalities show subliminal priming effects more for words rather than non-words, and both also show little evidence of semantic priming. They also dis- cuss how real-world examples of subliminal sound, such as those ‘so-called subliminal audiotapes, which supposedly convey relaxing subliminal audio messages’ (p. 617) have tended to fail as a result of unsatisfactory masking. The benefits accrued from these methods of delivery are more likely to be the result of supraliminal priming rather than subliminal priming. Therefore Kouider and Dupoux (2005) have provided an outline for future, albeit limited, adventures into subliminal audio presenta- tion. Watch out, the trolls might be coming after all . . .
Source: Kouider, S., & Dupoux, E. (2005). Subliminal speech prim-
on (i.e., very bright), then the light intensity values range from 0 to 7. For every light intensity value we presented we have data showing whether the participant saw the light being on or off. On the basis of these data (i.e., these perceptual reports) we can plot a psycho- physical function of the sort shown in Figure 4.6b).
So for every value of the stimulus intensity tested, we plot the probability that the participant actually reported seeing the stimulus. The problem now is to estimate the perceptual threshold, and this is where we have essentially a free hand. On the understanding that there is no useful sense in which an absolute threshold can be derived, we often end up choosing an arbitrary point as the threshold. Figure 4.6b shows a 50 per cent threshold. What this means is that, for the given level of stimulus energy, the participants will report seeing it on half the times it is presented. More typically though, a 75 per cent correct threshold is chosen. This is the point at which the participant is correct on 3/4 of the trials. Such a threshold can then be converted into a value on the actual light intensity dimension, for example, candelas/metre2(cd/m2). In
our light detection experiment what this means is that over the complete set of trials when the light was presented, the participant detected the light correctly on 75 per cent of occasions when the intensity was of a particular value estimated from our psychometric function. The particular intensity value signifies the 75 per cent correct threshold.
By now it should be clear that this rather detailed discussion of thresholds is leading towards considera- tion of how best to draw the line between non-conscious (or perhaps more fittingly pre-conscious) and consci- ous, between being unaware of the stimulus and being aware of the stimulus? More worrying, though, is that the 75 per cent correct threshold may be of little use in helping us draw a distinction between pre-conscious and conscious processing because on some trials the participant will be aware of the stimulus that falls below this threshold.
Clearly there are many methodological issues being side-stepped – such as how best to present the lights and what judgement we might ask of the participant – but most importantly we have failed to consider in any detail the psychological state of the participant. We have taken the participant to be essentially a passive sensor of stimulus information (the feature detector either comes on or it stays off and this is fully deter- mined by the amount of energy present) and what we have failed to recognise is that the participant is a decision-maker and now there is a clear difference with the simple feature detector. There is no sense in
Figure 4.6 Graphical representations of two sorts of thresholds
(a) An example of step function associated with an absolute threshold. The perception of the stimulus only takes place once the stimulus strength has exceeded a certain threshold value. The stimulus is either present or it is absent and there is no in-between. (b) A continuous monotonically increasing function with an arbitrary threshold applied. Evidence in favour of a stimulus increases in a smooth fashion but the person is only prepared to respond that the stimulus is present once the stimulus strength exceeds the threshold value.
Source: Haber, R. N., & Hershenson, M. (1979). The psychology of visual perception (fig. 5.1, p. 89). London: Holt, Rinehart
and Winston. Reproduced with permission from Ralph Haber.
After every stimulus presentation the participant was asked to report anything they thought they saw. The luminance of the target field was incremented slightly after every trial and the same target word was repeat- edly presented over a sequence of 18 trials. Now a clas- sic perceptual defence effect in these circumstances would be that participants would take longer to report taboo words correctly (such as ‘raped’ and ‘filth’) than neutral words (such as ‘mixer’ and ‘clove’). That is, neutral words could be ‘seen’ at lower levels of lumin- ance than could the taboo words – taboo words had higher perceptual thresholds than neutral words.
Taken at face value, such a result does have quite important consequences for understanding perception because perceptual defence can be construed as being an example of semantic activation without conscious identification. How could the perceptual threshold of the taboo and neutral masked words be differentially affected unless the meaning of the words had been accessed? In this regard perceptual defence poses dif- ficulties for any simple sequential stage theory of per- ceptual processing in which stimulus encoding must run to completion before any processes concerning the interpretation/identification of the stimulus takes place. The existence of perceptual defence seems to suggest that the meaning of the stimulus is affecting sensory encoding!
Given these sorts of considerations, there ensued much discussion over the claim that perceptual defence reflects a tendency to resist recognising threatening stimuli (Postman et al., 1953). Without doubt, this is quite an alarming thing to claim: do we really want a perceptual system that insulates us from threatening stimuli? If you are about to be hit by a speeding car, you need to know about this pretty quickly in order to try to avoid it! Maybe a more sensible approach is to argue that perceptual defence effects show that anxiety- provoking stimuli are detected, but at an unconscious level. This act of recognition then pre-empts anxiety
For example . . .
associated with environment tobacco smoke. Clearly if we can’t detect (either by smell or by sight) tobacco in the air because it is being masked by these addi- tives, then we will be more likely to think we’re in an environment free from the effects of second- hand smoke inhalation. So while we’re discussing these apparently theoretical issues, they do have real- life and potentially life-threatening consequences. If all of this is starting to feel a little remote, then the
work of Connolly, Wayne, Lymperis and Doherty (2000) should act as a sobering reminder of how important issues concerning masking and percep- tual thresholds can be. In their review, Connolly et al. (2000) discussed the efforts of tobacco companies who are introducing additives to cigarettes. The inten- tion is that additives should mask the sensations
which it decides when to come on! Given this, we need to consider how the psychological state of the partici- pant may affect the setting of a threshold. How might the participant’s motivations and expectations affect the difference between being aware of the stimulus and being unaware of it?
supra-threshold Above threshold.
sub-threshold Below threshold – subliminal.
subliminal perception Perception thought to occur below the threshold of consciousness.
stimulus factors Dimensions upon which a stimulus may vary.
subjective factors Factors that are inherent to participants – indicative of individual differences.
psychophysical function A graphical/mathematical means for relating a participant’s responses to some variation in a stimulus factor.