[1] New or Not - Recognition memory for Abstract Items

Background

This study was based around the Sternberg Paradigm for researching short-term recognition memory. Its primary goal was to investigate the 'extralist feature effect' where novel or surprising items are particularly easy to recognise as 'new' or not previously seen. There are many competing theories of why this effect occurs, but some posit that unusual features are particularly attention grabbing, allowing similarities between the test item and memory to be more easily ignored/filtered out.

Research Questions / Hypotheses

We aimed to manipulate the magnitude of this 'extralist feature effect' by varying the similarity between items on each memory list, and how the probe item differed from the rest. We predicted that increased variability on the list should make the 'attention grabbing' effect to be reduced. Additionally, this theory should result in items which are slightly dissimilar from the memory set in multiple ways to be harder to detect than a large dissimilarity in one way.

Participants

144 participants completed the study, with 30 excluded for low performance (inability to discriminate target from lure items). However, we will also run trial-level exclusions based on response times (too slow or fast).

Methods

In each session, participants saw 200 trials consisting of 3 study items that were abstract 'boxcar' rectangles which could vary in height, saturation and bar position. Participants were asked to memorise these three items, then decide if a fourth 'probe' item was one of the previous 3. Targets made up half or the probes, and were one of the studied items. Lures were produced by starting with a study item, and varying one or two properties by either 1 or 2 'just noticeable distances'.

Results

Analyses are incomplete, however the basic plan is to analyse whether the list-variability, number of varying dimensions, and distribution of dissimilarity across lure dimensions affects both accuracy and response time for recognition. We expect that increased variability in the study list will decrease sensitivity to small changes in the probe item, and that the effect of equal/unequal variability across dimensions will interact with list variability.

Implications

These results are planned for communication in both conferences and a journal article. Our findings will have implications for the processes of rapid recognition decisions, and contribute to our understanding of how humans are able to make fast, accurate recognition judgments