The suggestion by Mack & Rock (see Mack, Tang, Tuma, Kahn & Rock, 1992; Rock, Linnett, Grant & Mack, 1992; Mack & Rock, 1998) that earlier investigations of so-called ‘pre-attentive’ processing without attention, may not have controlled for diffuse attention across the visual scene, nor for the intentional goals o f the observer.
are well taken. On the other hand there are a number o f important potential flaws with the ‘Inattentional Blindness’ (IB) methodology they have introduced.
Relying on post-display surprise questioning is inherently problematic for studies of this type. Subjects could have processed the background grouping on-line, in the IB experiments, and yet not recall the information later due to failures of explicit memory (e.g. see Wolfe, 1999). Alternatively the critical grouping processes could operate at a preconscious level, so that during initial presentation any grouping does not reach awareness but is nevertheless encoded (e.g. see, Moore, 2001). Additionally, subjects could possess very low-confidence knowledge of the background grouping, yet with some versions o f Mack et al’s methodology this might not be identified, as subjects were not always pushed to make forced-choices between distinct alternatives.
It may be more appropriate to measure any grouping that might arise under
conditions of inattention in an indirect and implicit manner (to eliminate the problem
of low-confidence or unconscious knowledge), using on-line measures (to eliminate
the issue of memory problems). The processing measured in this way may clearly differ from that assessed with explicit surprise retrospective questioning, as used by Mack, Rock and their colleagues. Subjects often remain unaware of the properties that affect their behaviour implicitly, yet it still remains vital to assess this processing (see Merickle, Smilek & Eastwood (2001), for a recent review o f implicit visual processing without explicit awareness).
Mack and Rock themselves attempted to address the criticisms involving the role of memorial demands, in a number o f experiments (Mack & Rock, 1998). Within these tasks, subjects completed the cross task parafoveally (as in the experiment discussed earlier within this chapter) while a small rectangle appeared at the fovea.
Immediately after the offset of the cross and original rectangle, a second rectangle appeared. This was identical to the first except that it was shifted slightly in location. Mack and Rock reasoned that if the first item is fleetingly perceived but then forgotten it might nevertheless produce apparent motion attributed to the second rectangle. They believed that this was possible as Weisenfelder and Blake (1991) had demonstrated apparent-motion perception despite one of the inducing stimuli being
suppressed due to binocular rivalry. However, most of Mack and Rock’s subjects
failed to perceive this apparent motion (in fact many did not even perceive the second rectangle). Mack and Rock concluded that this was clear evidence that stimuli to which subjects are inattentively blind are not fleetingly perceived. However, the fact that many subjects did not perceive the second rectangle suggests only that is possible that apparent motion cannot be consciously perceived if both stimuli are unattended. If this were the case, it would render the method unsuitable for investigating the issue at question. Additionally, Mack and Rock used these experiments to argue only against one criticism of their original studies (that subjects could have forgotten about the experience before being questioned). Their experiments did not address the problems with using only direct, explicit methods to examine any implicit processing.
Moore and Egeth (1997) attempted to devise experiments that sought to
measure grouping processes implicitly, under conditions of inattention. Subjects
judged the length of two horizontal lines presented one above the other. The
background consisted of dots that, if grouped by common contrast polarity, should induce the standard Muller-Lyer and Ponzo illusions for the extent of each horizontal line (see Figure 1.4 for an example stimulus). Results demonstrated that subjects’ judgements of line length were biased by the geometric illusion, which was taken to suggest inattentive organisation of the background formation. This was despite the
fact that subjects retrospectively reported little or no knowledge of the background organisation, on surprise retrospective questioning of the type used by Mack & Rock (c.f. Mack, Tang, Tuma, Kahn and Rock, 1992; Rock, Linnet, Grant & Mack, 1992; Mack and Rock, 1998).
Figure 1.4.
Schematic representation of a sample trial from M oore & Egeth (1997). Subjects Judged which of the two
horizontal lines was longest. Any
implicit b ackground grouping should bias responses to the top line in this example, as the dots g roup to produce the lengthening version of the Muller- Lyer illusion for th at line.
Moore and Egeth’s (1997) study provides a conceptual advance on standard measurements of inattentive blindness, due to the indirect ‘implicit’ on-line nature of their measure. But unfortunately their particular stimuli leave open the possibility that their results are due to processes other than fully inattentive grouping. For example, it should ideally be the case that ‘inattentively grouped’ items are as separate from, and as different to, the primary task as possible, thereby preserving Rock and Mack’s criteria for preventing dispersal of diffuse attention. Recently it has been shown that the more similar a task-irrelevant item is to task-relevant stimuli, the more likely it is to be noticed even within a standard IB paradigm (Most, Simons, Scholl, Jimenez, Clifford, & Chabris, 2001), presumably due to attention spreading easily to these items. The critical background stimuli in Moore & Egeth’s study were the same
colour and contrast as the lines, so even though the background pattern was not accurately reported retrospectively, it is possible that it was not entirely ‘inattended'.
Moreover, the identically coloured dots and line were very close to each other, and the dots themselves were close together. The closeness o f the inducing dots to the target lines (only 0.62 degrees of a visual angle away) attracts criticism based on the role of location in IB paradigms (e.g. see Most, Simons, Scholl & Chabris, 2000). Moreover, due to the closeness of the dots to each other and their particular contrast polarities, it seems possible that rather than true ‘grouping’ by similarity, mere blurring at low spatial frequencies could create an elongated ‘line’ connecting the black dots, as the contrast between them and the surrounding white dots is so strong, and neighbouring dots very close. Finally, in Moore & Egeth’s study, attention was drawn in initial instructions to the black and white background dots. This is best avoided in studies hoping to measure inattentive processes.
Smith and Merikle (1999) also sought to investigate implicit processing within an ‘Inattentional Blindness’ paradigm. Subjects were presented with the standard cross stimulus used by Mack and Rock (e.g.. Mack, Tang, Tuma, Kahn & Rock, 1992) and a word on each trial. On some trials subjects were instructed to attend to the cross and on other trials to the words. They examined the difference in processing of attended versus unattended words by introducing a stem-completion task at the end of the cross-and-word displays. For this, subjects were required to complete the three-
letter word stems with the first word that came to mind except words that had
appeared in the previous displays (c.f. Debner & Jacoby, 1994). Results demonstrated that for attended words, subjects successfully completed the task by entering words that had not featured earlier. However, for unattended words, subjects consistently completed the stems with the words that had been earlier presented. This is clear
evidence of priming from unattended words in an ‘Inattentional Blindness’ paradigm. Moreover, Merikle, Smilek and Eastwood (2001) emphasise that unattended information could not be prevented from influencing the stem-completion task under the ‘exclusion’ conditions, unlike the effect from attended words.
An additional paradigm that has been used to demonstrate implicit processing in the absence of explicit detection is the ‘attentional blink’ phenomenon. Within this method subjects typically fail to detect a second target appearing in an RSVP display if it appears within approximately 100 to 400 msecs of a first target (e.g. see Raymond, Shapiro & Amell, 1992; Shapiro & Raymond, 1994; Shapiro, Raymond & Amell, 1994). Results from these experiments have been used to argue that attention cannot be immediately re-deployed after successful detection o f a target. Thus, failures to detect the second target may be analogous in some respects to the other failures to explicitly detect stimuli without attention described earlier in this chapter. More recently the method has been used to demonstrate that missed items occurring in the ‘blinked’ period can in fact be implicitly processed. In a study by Shapiro, Driver, Ward and Sorensen (1997), responses to a third target word were facilitated if it was related to an undetected second target word, despite the fact that the second target had not been explicitly detected. Thus, the meaning of the undetected second target appeared to be extracted and prime detection of a related third target. Additionally, a study completed by Luck, Vogel and Shapiro (1996) also presented words during the
‘blinked’ period. Consistent with previous findings of the ‘attentional blink’ and
harmonious with the ‘Inattentional Blindness’ literature, subjects could not recall the
words presented during the ‘blinked’ period. However, Luck et al utilised
these unattended words had been sufficiently extracted to affect the N400 wave of the ERP.