The impact of fillers on lineup performance

Abstract

Filler siphoning theory posits that the presence of fillers (known innocents) in a lineup protects an innocent suspect from being chosen by siphoning choices away from that innocent suspect. This mechanism has been proposed as an explanation for why simultaneous lineups (viewing all lineup members at once) induces better performance than showups (one-person identification procedures). We implemented filler siphoning in a computational model (WITNESS, Clark, Applied Cognitive Psychology 17:629–654, 2003), and explored the impact of the number of fillers (lineup size) and filler quality on simultaneous and sequential lineups (viewing lineups members in sequence), and compared both to showups. In limited situations, we found that filler siphoning can produce a simultaneous lineup performance advantage, but one that is insufficient in magnitude to explain empirical data. However, the magnitude of the empirical simultaneous lineup advantage can be approximated once criterial variability is added to the model. But this modification works by negatively impacting showups rather than promoting more filler siphoning. In sequential lineups, fillers were found to harm performance. Filler siphoning fails to clarify the relationship between simultaneous lineups and sequential lineups or showups. By incorporating constructs like filler siphoning and criterial variability into a computational model, and trying to approximate empirical data, we can sort through explanations of eyewitness decision-making, a prerequisite for policy recommendations.Electronic supplementary materialThe online version of this article (doi:10.1186/s41235-017-0084-1) contains supplementary material, which is available to authorized users.

The number of fillers may not matter as long as they all match the description: The effect of simultaneous lineup size on eyewitness identification

“…Again, we will not compare these models of theoretical discriminability in the present study due to our focus on empirical discriminability and resulting policy recommendations. Rather, we will be the first to test a lineup size prediction from DFD with empirical data, in an attempt to support the simulations by Wetmore et al (). Moreover, we will determine how lineup size influences another relevant issue: the confidence‐accuracy (CA) relationship.…”

“…This is of utmost importance for recommending a particular procedure to policy makers (Mickes, ; Wixted & Mickes, ). Rather than involving empirical discriminability, DFS is a theory of response bias concerning differences in whether a filler, a suspect, or neither are chosen (Colloff, Wade, Strange, & Wixted, ; Rotello & Chen, ; Wetmore, McAdoo, Gronlund, & Neuschatz, ). Moreover, evidence that filler siphoning affects discriminability has not been borne out by recent research (Colloff & Wixted, ; Rotello & Chen, ; Wetmore et al, ).…”

“…Rather than involving empirical discriminability, DFS is a theory of response bias concerning differences in whether a filler, a suspect, or neither are chosen (Colloff, Wade, Strange, & Wixted, ; Rotello & Chen, ; Wetmore, McAdoo, Gronlund, & Neuschatz, ). Moreover, evidence that filler siphoning affects discriminability has not been borne out by recent research (Colloff & Wixted, ; Rotello & Chen, ; Wetmore et al, ). Based in part on these reasons, our primary focus is testing an extension of the DFD hypothesis, which has more explicit predictions concerning empirical discriminability.…”

“…Sequential lineup presentation may also lead witnesses to compare the faces, but similarities and differences across members will not be as readily obvious nor noticed. Additionally, sequential lineup performance depends on an interplay of other factors (e.g., suspect position; Gronlund et al, ), and there is some evidence that increasing sequential lineup size actually reduces discriminability (Wetmore et al, ). This may not occur when adding fillers to simultaneous lineups though, because DFD suggests that there are memory benefits associated with comparing faces that are side‐by‐side.…”

“…Wetmore et al () ran simulations using ROCs to compare performance between showups and varying simultaneous lineup sizes (3‐, 6‐, 12‐members) and found no advantage of increasing fillers beyond a fair 3‐member lineup. In their simulations, they used a Best Above Criterion decision rule, such that the face considered the best match to memory for the target's face is chosen, given that it is above a set criterion.…”