Animacy effects extend to working memory: results from serial order recall tasks

ABSTRACT

Words that represent animate entities (e.g., dog) are recalled from long-term memory more accurately than words representing inanimate entities (e.g., pan). In this research, we examined whether the animacy effect extends to working memory (WM). The potential roles of WM maintenance strategies (rehearsal and directed attention) were also examined. Participants performed serial order recall tasks with sets of 3, 4, 5 or 6 words that were either animate or inanimate. In stage 1 of Experiment 1, participants received no instructions regarding how to maintain the words. In stage 2, participants received either no instructions, sub-vocal rehearsal or directed attention instructions. In all instruction groups and at both stages, significant animacy effects were observed at set sizes of 4, 5 and 6 but not set size 3. The animacy effect was larger at set size 5 than other set sizes. In Experiment 2, participants completed the task with and without articulatory suppression. In both conditions, animacy effects were observed at set sizes 4, 5 and 6, but not at set size 3. The magnitude of the animacy effect declined as set size increased from 4 to 6. Animacy appears to enhance serial recall, but the effect is not dependent on rehearsal.

KEYWORDS:

• Animacy effect
• working memory
• serial order recall
• functional evolutionary theory
• articulatory suppression

Data availability statement

The data from this study cannot be shared due to ethical clearance restrictions only allowing the research team access to the data.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Glenda Andrews http://orcid.org/0000-0001-8047-9151

Notes

1 A sequential increase in set size for WM span tasks is commonly used (Conway et al., 2005) and is suitable when there is no intention to predict higher order cognitive performance (St Clair-Thompson, 2012). Serial recall accuracy does not differ significantly when set sizes are presented in random order versus sequentially increasing order (St Clair-Thompson, 2012). Similarly, no differences in accuracy are observed when participants are aware of what the set size will be compared to when set sizes are presented in a random and unpredictable order (Grenfell-Essam & Ward, 2012).

2 In addition to the two measures mentioned, proportion of lists correctly recalled and maximal span were also calculated. Proportion of lists recalled awards credit for lists correctly recalled in their entirety, while maximal span was calculated according to Cowan et al. (1994). The results of all four measures of serial recall accuracy were generally consistent, indicating that the animacy effects reported in this study are not an artefact of the scoring method used.

3 The results of Wilcoxon Signed Ranks tests indicated that animate words were not recalled with greater accuracy than inanimate words at set size 3 at either stage 1 (p = .078) or stage 2 (p = .389). This was also the case for Experiment 2, where there was no significant effect of animacy at set size 3 in either the suppression (p = .107) or the no suppression (p = .112) condition.

4 Single sample t-tests were also conducted to compare cumulative spans to a value of 5 given Cowan’s estimate of WM capacity being 3–5 chunks. At both stage 1, t (94) = −6.01, p < .001 and stage 2, t (94) = −3.47, p = .001, cumulative span for animate words was significantly less than 5.

5 The Hebb repetition effect is the observation that if the same sequence of stimuli is repeated every few trials, recall of this sequence improves over successive presentations (Hebb, 1961; Hurlstone et al., 2014).

Additional information

Funding

This research was supported in part by an Australian Government Research Training Program Stipend to Matthew J. Daley by the Australian Government.