Abstract
Estimates of visual working memory capacity (K) vary widely across individuals (Vogel & Awh, 2008), and correlate with measures of intellectual ability. However, typical measures of K are potentially confounded with attentional lapses. It is unclear whether incorrect responses are due limitations in capacity, or a complete lapse of attention that results in no items being stored in working memory. Recently, methods using multiple set sizes and new analysis techniques (Rouder, Morey, Cowan, Zwilling, Morey, & Pratte, 2008; Morey, 2011) were introduced to dissociate attentional lapses and K. However, there is evidence that participants may exert reduced effort at larger array sizes, leading to a decreased estimate of K (Rouder, et al., 2008). Here, we evaluate a new method of assessing K-called Multiple Change Detection (MCD)-that provides concurrent measures of attentional lapses and K while eliminating differential effort allocation across set sizes (Gibson, Wasserman, & Luck, 2011). Rather than varying the set size and changing a single item, this method keeps set size constant and varies the number of items that change. The number of changes is not known in advance, so observers cannot strategically vary their level of effort.
To evaluate MCD, we performed Monte Carlo simulations contrasting MCD and traditional methods for estimating K. MCD-based estimates of K were less distorted by attentional lapses. We applied this new method to people with schizophrenia (PSZ) and healthy control subjects (HCS) to determine whether previous reports of reduced K in PSZ were an artifact of an increased lapse rate. We found that PSZ had more frequent attentional lapses than HCS, but capacity was still reduced in PSZ. By providing separate estimates of attentional lapses and capacity and eliminating the possibility of differential effort allocation across set sizes, MCD provides a promising new method for estimating visual working memory capacity.
Meeting abstract presented at VSS 2013