Report About the Attentional Blink: Based on the Research Conducted by Zivony and Lamy

Report about the Attentional Blink: Based on the Research Conducted by Zivony and Lamy

When two targets (T1 and T2) are presented within 200ms to 500ms from each other in rapid visual presentation (RSVP), identification of T2 is often impaired. This phenomenon referred to as the attentional blink (AB). The essential paradigm of attentional blink is the RSVP paradigm proposed by Chun and Potter (1995). The stimulus sequence displays successively at the speed of 6-20 items (letter, digit, word, picture, etc.) per second and participants are asked to distinguish or identify the first target (T1) and the second target (T2) which appears in the 1-8 lag of T1. As the development of the research, some variants occur, mainly including two categories: RSVP paradigm with one stimulus sequence and RSVP paradigm with multiple stimulus sequences. The factors such as task difficulty, target property, spatial factor, temporal factor, cognitive state can influence the results.

In this literature, depending on whether they assume the blink to prevent T2 from triggering an attentional episode(the stage initiated whenever a target matched target template is detected from the RSVP stream), the authors divided the theories about AB after two-stage model into two groups: disrupted-control theories and disrupted-engagement theories. Disrupted-control theories suggested that while T2 may be processed up to the semantic level, target detection doesn’t occur during the blink due to a failure of attentional control and as a consequence, T2 is not selected for report. Although there’re some differences between different theories, they all share the assumptions that AB disrupts the processes responsible for matching incoming information to the attentional set and sparing of a cued T2 occurs because the cue reinstates such target-detection process. Disrupted-engagement theories suggested that AB doesn’t impair target-detection processes, but impair the attentional enhancement (i.e. engagement) of T2 that is responsible for granting the information extracted from T2 access to working memory and consciousness. There theories share some premises. First, they assume that during AB, target-defining features are detected and matched with the attention set. Second, they suggest that the appearance of a target-defining feature automatically triggers an attentional episode. Finally, during the processing of T1, redeployment of attentional engagement is not deducted but effectively delayed. Consequently, although T2 initiates an attentional episode, attentional engagement is not so fast enough for the T2 to be selected and the item following it is selected instead.

Experiment 1

The design of EXP1 is based on a core difference between the two types of theories: disrupted-engagement theories predict that following a relevant-color cue, attention will be involuntarily configured, thus any item following this cue should be selected, no matter whether it is a target or a distractor. However, disrupted-control theories predict that, as the cue reactive target detection processes, attention will be deployed only to a target that follows it.

The paradigm used in this experiment was adapted from the paradigm developed by Folk and colleagues (2009) in which the blink occurs following involuntary attentional capture by a distractor rather than following selection of a first target. They found that AB can occurs, when the color-defined target (letter) following a nontarget-colored letter which is enclosed by a target-colored outline square distractor.

In this experiment, authors used two distractors (D1 and D2) and one target. Each distractor could be either surround by a relevant-color (the same as the target) circle or an irrelevant-color. Both disrupted-engagement and disrupted-control theories predict that the relevant-color D1 should capture attention and relevant-color D2 should therefore fall within the blink. However, the predictions of D2+1 are different. According to the disrupted-engagement theories, D2 should trigger an attentional episode when it’s in the relevant color but the engagement is delayed, so the D2+1 should be selected even if it’s not a target. Reversely, according to disrupted-control theories, a relevant-color D2 should reactivate target detection processes so that only target should be selected. So within blink, disrupted-engagement theories predict the more D2+1 intrusions when D2 is in relevant-color than in the irrelevant-color. By contrast, disrupt-control theories predict there’s no difference between these two conditions.

Method

The RSVP stream consisted of 15 frames, each letter, which were selected randomly, containing a letter enclosed in an outline shape. Each frame displayed 100ms. The EXP1 was designed as a 2(target in green color or in red color) ×2 (D1 in relevant-color or irrelevant-color) ×2 (D1 appeared in the third condition or ninth condition) ×2(D2 in relevant-color or irrelevant-color) × D2 compatibility mixed design. The target color is the between-subject variable, while others are within-subject variables. The accuracy of target identification was tested as the dependent variable. D2 always followed D1 by two frames and target followed D2 by two frames.  Participants were asked to report the target letter with no time pressure. All the conditions had equal chance to appear and appeared randomly.

A typical process shown below.

[pic 1]

Results

Attentional blink

After analysis of ANOVA, only when D2 was in the irrelevant, the rate of target identification was lower following a relevant- than an irrelevant-color D1.

D2 compatibility effect

The compatibility between D2 and the target letter had a weaker effect inside than outside the blink. In this experiment, the compatibility effect was largest in the irrelevant-color D1 & relevant-color D2 condition and did not differ between the other three conditions.

D2+1 identity intrusions

The intrusion was operationally defined as the mean proportion of error trials in which D2+1 was reported instead of the target. The analysis indicated that the intrusions following the relevant-color D2 were equally high whether D2 in or not in the blink. In contrast, when D2 in irrelevant-color, the intrusions were higher when D2 inside the blink. In addition, when D2 in the blink, the intrusions were more frequent following a relevant-color D2.

Discussion

EXP1 yielded three main findings. First, analysis of D2-target compatibility effects confirmed that the relevant-color D1 produced an attentional blink. Second, regardless of D1 color, D2+1 intrusions were more frequent following a relevant-color D2, which is in line with the prediction made by disrupted-engagement theories. Finally, an unexpected finding is that D2+1 intrusions following an irrelevant-color D2 were more frequent when D2 appeared outside the blink.