• No results found

The two experiments described within this chapter provide converging evidence that some visual grouping processes may operate without attention (at least, under conditions of “inattention” as defined by Mack & Rock). Experiment 1 demonstrated that subjects’ performance in the matrix task was influenced by change or continuity in the vertical versus horizontal grouping o f background dots, on the

basis o f common colour, when performing the central matrix task. Responses,

particularly to matrix patterns that changed across the two displays, were facilitated by congruent change in the background configuration. This is despite the fact that subjects exhibited little or no explicit knowledge o f exact background organisation, when tested in the Mack & Rock fashion with surprise explicit questions. These results were extended in Experiment 2. This experiment showed that the previous results were not specific only to particular types o f highly ordered grouping, as subjects again demonstrated a congruency effect within responses, despite the fact that background circles could now be grouped into either vertical columns, or into a

pseudo-random pattern with no overall ordered colour configuration. Again this

pattern was primarily found for responses to changing matrices. In this second

experiment, subjects overall remained unable to correctly answer explicit questions about the background grouping after inattentive presentation.

Experiment 2 also counterbalanced the colour changes that occurred within each trial. Every circle now changed colour during each trial regardless of whether overall grouping alters; this was done in order to prevent any apparent grouping effect being actually based on just local change detection in a few or only one circle, rather than processing of the entire background grouping. Previously all subjects were

presented with red circles changing to blue and green circles changing to yellow, whereas in Experiment 2 half the subjects had this original colour mapping and the other received the opposite; red changing to yellow and green to blue. Results were not affected by this manipulation, confirming that the implicit effect in Experiment 1 was not the result of an exceptionally salient local colour-change specific to one type of colour mapping change.

Overall, the results of these two experiments imply that previous evidence suggesting that there is no visual segmentation of the scene without attention (Mack, Tang, Tuma, Kahn and Rock, 1992; Rock, Linnet, Grant & Mack, 1992) may have been over interpreted, as the authors did not take into account possible implicit processing. It seems that implicit processing of background grouping may in fact take place, as in order to produce the congruency effects seen in these experiments, the visual system presumably must extract and process the grouping o f the background circles in both displays of each trial to some extent. This is in spite of the fact that, using Mack & Rock’s criteria alone, it would seem on the basis o f the explicit surprise questions that no processing of the organisation of the background circles had taken place, as subjects were at chance when explicitly judging the grouped configuration of the previous display.

There are several outstanding issues still to be clarified. The results thus far showed more consistent congruency effects for responses to changing matrices. Identifying whether a pattern remains the same appeared to be easier, as shown by the main effect o f matrix type in error-rate data for both Experiments 1 and 2. During both experiments, responses were significantly more accurate to ‘same’ matrices, i.e., those that did not change over the course of the two successive displays that

comprised each trial. Responses to ‘same’ matrices may thus be operating at a near ceiling level, which could leave them less influenced by the background condition.

Additionally, results from the explicit question block did vary somewhat

between the two experiments. For Experiment 2 the pattern o f correct responses

matches those discovered by Mack and Rock throughout their demonstrations of so- called “Inattentional Blindness” (Mack, Tang, Tuma, Kahn and Rock, 1992; Rock, Linnet, Grant & Mack, 1992; Mack & Rock, 1998). That is, chance performance for the inattentive questions, around chance for the divided attention questions and finally above-chance for the final control questions. However, in Experiment 1 there was a slightly different pattern. That is, the number of correct responses was above chance for the very first inattentive question. This is possibly explained by the nature of the patterns used in that experiment. The two alternative types o f grouping were always arranged ‘systematically’ so that statistically there were more systematically arranged grouping patterns in Experiment 1. This could enable subjects to better answer this question. This point will be returned to in Chapter 4 when the same grouping patterns are again used. Alternatively, this result could simply be the result of a response bias

within the subjects participating in Experiment 1. Importantly, these subjects were

not above chance when answering the second inattentive question. Moreover, the critical congruency effect was not found more reliably in those who answered the inattentive questions correctly in Experiment 1.

However, within Experiment 2 there was some evidence that subjects who were correct at the answering the second inattentive question (“Were the background circles grouped into vertical columns or arranged into a pattern with no overall grouping?”) showed a slightly stronger interaction between matrix identity and background configuration condition, although this interaction also approached

reliability even for the incorrect group. Furthermore, both groups demonstrated the critical congruency effect in that both were more accurate to respond to different matrices when the background organisation also changed.

The evidence from these two studies suggests some preserved processing of background grouping (by common colour) under inattentive conditions. Mack and Rock’s criteria for assessing inattentive processing would suggest, on the basis of the explicit questions, that the background grouping had not been perceived. However, the new indirect on-line test in the matrix task reveals a consistent congruency effect, when the change to the target in the central task is compatible with changed grouping

in the background configuration. It appears that the perception of visual grouping

without attention may have been underestimated by previous direct questioning methods in the “Inattentional Blindness” tradition (Mack, Tang, Tuma, Kahn and Rock, 1992; Rock, Linnet, Grant & Mack, 1992; Mack & Rock, 1998).

Chapter 3

Improved Indirect Measures of Background