It’s your choice! – Or is it really?

We recently addressed this question in a study (Sagana, Sauerland, & Merckelbach, 2012, Experiment 1). Participants watched four video fragments, each showing a mock crime. Four different actors were involved in each fragment: a perpetrator, a victim, and two bystanders. After each fragment, participants were asked to identify each of the four actors from four different lineups (they were not given the option to reject a lineup). In total, participants had to make 16 identification decisions. After each decision there was a short delay before participants were presented with their choice again and asked to explain their decision. At this point, however, some of the participants’ choices were manipulated so that they were presented with a non-chosen lineup member. To our surprise, 36 out of the 37 participants (97%) detected the manipulation. Furthermore, participants were highly accurate in their identification decisions in both manipulated and non-manipulated trials (both 73%).

In a follow-up experiment, we inserted a longer time interval between making the identification decision and being questioned. This not only created a more ecologically valid situation, but was also meant to increase the ambiguity of the situation. Specifically, participants were presented with the manipulated outcome 48 hours after their identification decision (Sagana et al., 2012, Experiment 4). Preliminary results from 30 participants showed that 21 (35%) out of 60 manipulated trials were not detected. Thus, choice blindness is a relevant factor in identification decisions, particularly when ambiguity is added to the task, resulting in possibly devastating consequences for the defendant and the judicial system.

Carry-Over Effects of Choice Blindness

Apart from the immediate impact of choice blindness, accumulating evidence indicates that this phenomenon may continue to influence decision making beyond the time of the initial manipulation. For example, in their face-preference study, Hall, Johansson, Tärning and Sikström (in preparation; as cited in Johansson et al., 2008) observed that when participants had to indicate their preference again in a second round directly following the first one, choice-blind participants were significantly more likely to select the manipulated face. In their study on symptom escalation and choice blindness, Merckelbach, Jelicic, and Pieters (2011a) found that blind participants were more likely to report increased symptom ratings at a re-test one week later. Similarly, Sauerland, Schell and colleagues (2012) noted that four weeks after the initial interview, blind participants tended to increase their ratings on how often they had committed certain norm-violating behaviors.

Choice blindness can also influence how confident participants are in their ratings. In our line-up experiments (Sagana et al., 2012, Experiments 1, 4), eyewitnesses had considerably less confidence in their identifications when their decision had been manipulated, compared to non-manipulated identifications. This was true even for eyewitnesses who had noticed the manipulation. That means that participants’ confidence ratings were deflated, even if they realized that a switch had taken place. Research on eyewitness identification shows that high confidence levels can be a reliable indicator of accuracy (Brewer & Wells, 2006; Sauerland, Sagana, & Sporer, in press; Sauerland & Sporer, 2009). The more certain witnesses feel about the identification, the higher are the chances that they are correct. However, if choice manipulations can alter confidence, then the linearity of the relation is destroyed and hence, confidence is no longer diagnostic of identification accuracy.

article author(s)