Our perception of event duration appears to be modulated by our recent sensory history. For example, the perceived duration of the first stimulus in a stream of identical stimuli is typically overestimated (Rose et al., 1995). A related effect concerns the perception of infrequent or unexpected “oddball”
62
stimuli (see Figure 33) whose perceived duration is expanded relative to that of expected or frequent “standard” stimuli (Tse et al., 2004). It was initially suggested that this “subjective time dilation” increased the perceived duration of oddballs by approximately 30-50% (Tse et al., 2004). However, subsequent studies have suggested that this figure was grossly overestimated (Seifried et al., 2010), revealing a more modest expansion of around 10% (Ulrich et al., 2006(a), Chen et al., 2009, Pariyadath et al., 2007, van Wassenhove et al., 2008). The effect seems to be most robust for stimuli that are expanding in size, i.e. looming or approaching (Tse et al., 2004, van Wassenhove et al., 2008, New et al., 2009) and can be eliminated (New et al., 2009) with contracting or receding oddballs. The effect is reduced (Tse et al., 2004) or reversed (van Wassenhove et al., 2008), when a static oddball is presented within a stream of expanding standards. This implies an ecological “alerting” function in which an organism may respond to a possible threat more quickly and is consistent with reports of time slowing down in threatening situations (Campbell et al., 2007, Stetson et al., 2007). Inconsistent with this explanation, however, is the fact that similar effects have been reported for stationary stimuli (Chen et al., 2009, Tse et al., 2004, Pariyadath et al., 2007).
63
Figure 33 The initial stimulus and “oddball” appear to be expanded in duration compared to other stimuli in the series. This may be due to their unexpected nature or it may be that the other stimuli in the series are showing a contraction perceived duration due to repetition suppression of neural firing rates (Pariyadath et al., 2008).
There are two main competing explanations of subjective time dilation in relation to the oddball effect. The arousal theory claims that the alerting effect of an oddball causes a central internal pacemaker (Creelman, 1962, Treisman, 1963) to speed up, resulting in a subjective prolongation of time (Seifried et al., 2010). Alternatively an increase in attention could lead to more pulses being counted. These explanations receive support from the finding that subjective time dilation is a global phenomenon affecting the whole visual field, not just the oddball or its immediate surround (New et al., 2009). The centralised arousal theory is, however, difficult to reconcile with several experimental results: multisensory versions of the subjective time
64
dilation show asymmetric transfer between senses (Chen et al., 2009, van Wassenhove et al., 2008); the expansion of perceived duration can be generated with oddballs that are entirely predictable (van Wassenhove et al., 2008) and the fact that supposedly ‘emotive’ stimuli do not result in a greater expansion of perceived time (Pariyadath et al., 2007).
The information processing theory (Tse et al., 2004), on the other hand, proposes that the rate at which information is processed acts as the pacemaker component of our timekeeping system. In other words, ‘bits’ of information act as a counter with which we estimate the passage of time. This model suggests that the additional processing resources brought to bear for novel stimuli increase the overall rate at which information is processed, and the greater number of bits processed per unit time leads to an expansion of perceived duration. A related model is the “coding efficiency” model (Eagleman et al., 2009) where perceived event duration is directly related to the neural resources expended during the event’s processing by the nervous system. In this model, repeated presentations of the expected or ‘standard’ stimulus leads to progressively more efficient encoding of this stimulus – a phenomenon termed repetition suppression (Grill-Spector et al., 2006, Henson et al., 2003) - such that, on re-appearance, reduced neural activity levels induce a perceived contraction in the duration of the standard, relative to the non-suppressed oddball stimulus. This same mechanism could also explain the “novelty” effect of Rose and Summers (1995).
The coding efficiency hypothesis arose following a series of experiments examining the oddball effect and the arousal explanation (Pariyadath et al., 2007). The authors increased the emotional impact of the oddballs in their
65
experiments by using emotionally charged images, such as growling dogs and spiders. They found that there was no increase in duration expansion with these oddballs when compared to the more neutral oddballs used in earlier experiments. This led them to conclude that if attention is the crucial element in the oddball effect it must either saturate at around 15% or the oddball effect is caused by some other mechanism. The authors propose that the latter is in fact the case. This suggestion is based the findings in the electrophysiology literature that repeated presentations of the same stimulus lead to a reduction of neural responses. Unexpected stimuli with non- suppressed firing rates would therefore appear longer by comparison (Pariyadath et al., 2007). However, since attention has been shown to increase the firing rate of cortical sensory neurons (Chik et al., 2009, McAdams et al., 1999, Moran et al., 1985) and also the coherence of neural firing (Doesburg et al., 2008, Fell et al., 2003) the two ideas need not be mutually exclusive. As attention to familiar stimuli wanes, neural firing rates may reduce, a new stimulus would then produce an increase in attention and an accompanying increase in firing rates. Using a variation of the flicker fusion paradigm, the authors sought to investigate the two hypotheses.
The experiment involved repeatedly presenting letters for very brief periods such that explicit temporal judgments were impossible. In some trials the letters were the same, whilst in others they were different. Because the letters were presented for very brief periods, more than one appeared to be on the screen at a time due to visual persistence.
Pariyadath and Eagleman found that fewer characters were perceived at a time during the repeated condition, due, they suggest, to a contraction of
66
visual persistence of the repeated stimuli (see Figure 34). They call this the proliferation effect and favour a neural suppression hypothesis because neural activity may be measured whereas as attention is more subjective and difficult to gauge. The authors suggest that a way to distinguish between attention and neural activity would be to conduct a range of experiments with repeated stimuli in which attention is manipulated by increasing the mental load of the experimental tasks (Pariyadath et al., 2008).
Figure 34 An experiment in which letters are presented to subjects for very brief durations. A) Shows sample sequences of stimulus presentations and the perceived numerosity when the letters are repeated and when they are different. B) Shows the number of letters perceived when they are repeated and when they are random. C) The authors suggest that the visual persistence of the repeated letters is reduced relative to the persistence of the random stimuli. D) When letters were presented for different physical durations whilst keeping the presentation frequency constant, no significant difference in perceived numerosity was found (Pariyadath et al., 2008).
67
Although arousal and information processing models offer appealing explanations as to why perceived duration is context dependent, a problem common to such models is an inability to explain the criterion adopted by the nervous system when deciding which events should be designated as ‘expected’ or ‘unexpected’. In other words, how ‘odd’ does an oddball stimulus need to be before its perceived duration is deemed to differ from its neighbours? The diverse nature of standard and oddball stimuli deployed makes inferences on this topic somewhat problematic. For example, oddballs have variously been defined by changes in geometric shape (Tse et al., 2004), stimulus size/intensity (Seifried et al., 2010, New et al., 2009, van Wassenhove et al., 2008), alphanumeric character and photographic image properties (Pariyadath et al., 2007), often altering multiple stimulus features simultaneously between oddball and standard trials. Although it has recently been proposed that high level factors play a role (Pariyadath et al., 2007), inferences as to the nature of this role are difficult without precise control over the stimulus parameters in question.