An investigation of cued recall of multiattribute stimuli.

Abstract

Memory performance for sequences of letters positioned in particular spatial locations in a 3 x 3 grid was examined by requiring participants to recall attributes of the target stimuli given 1 or 2 features of the stimuli as cues. Cuing asymmetry was observed between the serial-position curves of object and sequential-order information, and location and sequential-order information, when the stimuli were presented in both the same and different locations. After correcting for response bias, this asymmetry was attenuated for the stimuli presented in different locations and was eliminated for the stimuli presented in the same location. Contrary to the predictions of the fragmentation hypothesis (G. V. Jones, 1976), asymmetry was also observed between object and location information. The roles of spatial location and response bias are offered as explanations for previous contradictory claims for cuing symmetry between item and order information.

Response bias in visual serial order memory.

“…This is particularly true because serial position curves continue to provide important evidence that is used to make theoretical distinctions between memory stores and processes. As this and other articles have shown (Toglia & Kimble, 1976;Ward et al, 1997), the effects of response bias can seriously influence the functions of serial position curves. Therefore, unless the effects of response bias are considered, there is a danger that theoretical distinctions between memory stores and processes may be based on response decisions at output rather than knowledge of the events that occurred.…”

“…This pattern of results is illustrated in Figure IB. Moreover, two additional methods of correcting for response bias (see Toglia & Kimble, 1976;Ward, Churchill, & Musgrave, 1997), as well as measures of hits minus false alarms (Berch, 1975(Berch, ,1977, yielded similar patterns of results. This suggests that the response bias was responsible for the recency-to-primacy shift observed with this methodology.…”

“…This pattern of performance is illustrated in Figure 2 B. Again, the recency-to-primacy shift was eliminated using the proportion of certain responses as the dependent variable, and furthermore, the methods of correction proposed by Toglia and Kimble (1976) and Ward et al (1997) and of hits minus false alarms (Berch, 1975, 1977) produced similar results.…”

“…Unfortunately, the data reported by Korsnes et al (1996) and Korsnes and Magnussen (1996) do not allow such an examination of the distribution of responses. This is problematic, as it is known that the shape of the serial position curves can be greatly affected by response bias (e.g., Toglia & Kimble, 1976; Ward, Churchill, & Musgrave, 1997).…”

Modeling memory for absolute location.

“…For example, if one compares in the L | CAT matrix the near-miss errors for Location 2, in all three cases the errors nearer to the left-hand side, LI, are many fewer than are the errors to the center, L3 (6 compared with 37). Such centripetal bias is discussed in many previous studies (Huttenlocher et al, 1991;Nelson & Chaiklin, 1980;Taylor, 1961;Toglia & Kimble, 1976;Ward, Churchill, & Musgrave, 1997). However, the a posteriori probability of responding L3 in error is greater than LI, and any analysis of the distribution of errors needs to take this into account.…”