VENTAJAS DE LA PLANIFICACIÓN ESTRATÉGICA

In a standard ‘Change Blindness’ (CB) paradigm, observers typically fail (or take tens o f seconds) to detect substantial changes to a visual scene if a salient interruption intervenes, such as a transient flicker across the image (e.g., Rensink, O ’Regan & Clark, 1997). This is despite the fact that subjects are actively looking for a change and can move their eyes freely across the scene.

In fact researchers examining the perception of change across saccadic eye movements first uncovered this phenomenon, as people appear to be incredibly poor at detecting changes if these occur whilst they are making a saccadic eye movement, and the change does not concern the saccadic target (e.g., Li and Matin, 1990; Blackmore, Brelstaff, Nelson and Troscianko, 1995; McConkie and Currie, 1996; Currie, McConkie, Carlson-Radvansky and Irwin, 2000). Studies utilising a blank- screen interruption are viewed by some as being analogous to a transaccadic change, as a saccade will likewise produce a transient across the whole image and can also induce saccadic suppression. Additionally, studies comparing change detection across a blank interruption and across a saccade have found accuracy to be equally poor (e.g., Blackmore, Brelstaff, Nelson and Troscianko, 1995).

In the original CB experiment (Rensink, O’Regan & Clark, 1997) examining change detection across a blank screen interruption, rather than a saccade, a set of

full-colour photographs of everyday scenes were manipulated to create a secondary set, all featuring one considerable change from the original photograph (see Figure 5.1 for one example of their stimuli). For example, such changes included deleting an object from the scene or substantially changing the colour of an item. During the experiment, each original photograph and its changed partner were iterated several times a second. Importantly, interleaved between the photographs’ repetitions was a blank grey screen. This blank produces a flicker between the successive photographic images.

Observers viewed these flickering images and pressed a button when they spotted the change. Performance was poor: scenes were repeated many times before most changes were detected, in fact many were not noticed by the end of presentation time (1-minute). Interestingly, once subjects identify the change (or it is pointed out to them) it now seems trivially easy to see, the subjective appearance being that the disappearing item (in a deletion example) is now flashing on and off.

Rensink, O’Regan & Clark (1997, p.372) stated that “visual perception of change in an object occurs only when that object is given focused attention” (see also, Li and Matin, 1990; Blackmore, Brelstaff, Nelson and Troscianko, 1995; McConkie and Currie, 1996; Currie, McConkie, Carlson-Radvansky and Irwin, 2000). They suggested that visual changes in the environment usually produce a transient motion signal that exogenously cues attention towards the locus o f the change. With focused attention reaching the changing item, it is perceived with ease. However, flickering the image by introducing an interleaved blank screen causes a large transient motion signal that masks the smaller motion signals from the change itself. The result is that attention is not drawn to the relevant item and the change can only be detected by a slow ‘serial’ search across the entire image. Rensink et al’s (1997) results have been

used to suggest that internal representations of the visual world are extremely sparse. On this view, in contrast to our subjective experience of a richly textured visual environment, our representations consist only of a small amount of information at any one time, namely the information that is presently attended (e.g.. O’Regan,

1998).

Figure 5.1. Example of stimuli used by Rensink, O ’Regan & C la rk (1997).

Red outline highlights where change is occurring (the red outline was not present in the original study).

This is an example of an MI region of change.

The key evidence used by Rensink et al (1997) to support their claim for the role of attention in this paradigm is that detection times are faster for items described as being of ‘Central Interest’. They assert that these items endogenously attract attention more efficiently than other items as they are of greater interest to the observer. In order to define the regions and/or items of Central Interest (Cl) and Marginal Interest (MI) in the visual scenes, Rensink et al (1997; see also Rensink, 2000; O ’Regan, Deubel, Clark & Rensink, 2000) asked five independent observers to describe each photograph. Regions of Cl were defined as those that were mentioned by at least three of the panel whereas MI regions were those that were not mentioned by any of the five. Once this had been done, Rensink et al attempted to equate the Cl and MI areas roughly in terms of physical size and also magnitude of the change in

terms of intensity and colour. Regions of MI remained marginally larger on average. O ’Regan, Deubel, Clark & Rensink (2000) have described in more detail their recent attempts to equate the two types of scene area. For example, they have tried to match position of the regions in relation to the centre of the screen and mean proportion of modified pixels.

Despite attempting to equate physical salience to some extent, it remained the case that, typically, regions of Cl involved areas or objects that constituted the “main theme” (O’Regan, Deubel, Clark & Rensink, 2000) of the picture (e.g., Rensink, O’Regan & Clark, 1997; O’Regan, Deubel, Clark & Rensink, 2000; Rensink, 2000) unlike regions of MI. Thus, it was presumed that top-down processing would lead to preferential attention distribution to Cl regions. Therefore, advantages for change detection in Cl areas were taken to demonstrate (indirectly) the involvement of attentional systems in the perception of change within the paradigm.

Some apparent support for this comes from research into the standard eye movement paths of observers viewing visual scenes (e.g. Yarbus, 1967). These demonstrate that many portions of a picture are never fixated whereas other areas are repeatedly looked at. Those areas that are fixated many times are typically those that contain the central theme of the picture and thus would be defined as Cl regions by Rensink et al (1997). Evidence that attention and eye position generally works in harmony (e.g., Deubel & Schneider, 1996; Henderson, Pollatsek & Rayner, 1989) might then lend some support for the idea that areas of Cl will be preferentially attended.