Metacognitive processes accompanying the first stages of autobiographical retrieval in the self-memory system

ABSTRACT

According to Conway’s view, Autobiographical memory (AM) construction is accompanied by control processes. These processes range from filtering out relevant memories according to the current context, to generating or elaborating appropriate retrieval cues. These processes can be conceptualised as metacognition, the ability to control and monitor cognitive processes. Experimentally, little has been carried out to support the idea that metacognition is involved in AM. To assess this, we designed a task, the Feeling of Retrieval. Participants had to predict whether cue words would facilitate AM access (i.e., fluent access cues) or not (i.e., limited access cues) in a limited time (either 1 (Exp. 2) or 2 (Exp. 1) s). Later, they retrieved memories in response to both types of cues. Results show that cues judged as fluent access led to better AM generation, as illustrated by AM retrieval latency and a subjective measure of the ease with which the AMs were retrieved. These rapid predictions may rely on epistemic feelings and / or other mnemonic cues such as a partial retrieval of information. This metacognitive access to the earliest stages of AM retrieval illustrates the ability to monitor AM processes as proposed by Conway (2005).

KEYWORDS:

• Autobiographical memory
• retrieval mode
• metacognition
• feeling of fluency
• feeling of retrieval

Open Scholarship

This article has earned the Center for Open Science badges for Open Data, Open Materials and Preregistered. The data and materials are openly accessible at https://osf.io/r2j74/?view_only=d19767270c294be1b50b0c3bb53db43c, https://osf.io/r2j74/?view_only=d19767270c294be1b50b0c3bb53db43c and https://osf.io/r2j74/?view_only=d19767270c294be1b50b0c3bb53db43c.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

These studies were pre-registered in the Open Science Framework. All data, experiment scripts, and analysis scripts are available online (https://osf.io/r2j74/?view_only=d19767270c294be1b50b0c3bb53db43c).

Notes

1 We calculated a mean of RT for fluent access cues, and a mean of RT for limited access cues for each participant and created a score which reflected the difference in RT between fluent access cues and limited access cues for each participant: this is the variable we used to assess type of cue. The linear model contained this score (Wdiff) as its measure, and the reliance on successful retrieval (β1) as its predictor: Wdiff∼β1 (or Wdiff = β0 + β1 + ε).

2 We also started a third identical experiment, giving participants only 0.5 s to make the prediction judgement. We stopped recruiting participants after collecting data from 15 participants. Among these 15 participants, 9 failed the prediction task. That is, they failed to press the spacebar at least eight times so as to have eight words for which they would generate memories in the AM phase. Furthermore, all participants failed the attention checks and indicated that the presentation time was far too short to make a decision.

3 One participant’s response for this question was not been recorded.

4 Based on a reviewer’s suggestion, we analysed the effect of the type of cue on memory valence using a categorical approach. For each of our studies, we designed binomial generalised linear mixed models to inspect whether the type of cue would predict the valence of the memory. Memories associated with a negative rating (i.e., <0) were categorised as negative, those associated with a positive rating (i.e., >0) were categorised as positive. Those memories with a neutral rating (i.e., 0) were removed from the analysis (n = 163 (20%) in study 1, n = 94 (17%) in study 2). We found no effect of the type of cue on memory valence in both studies (respectively b = −0.1368, p = .481 in study 1; b = 0.1223, p = .566 in study 2).

5 We are grateful to the author for making this mean RT available to us.

6 In one of his latter keynote presentations in Grenoble, France, Conway (2017; a similar argument has been developed in Conway & Loveday, 2015) criticised the introspective approach towards studying autobiographical memory. He again stated that “ … the system elaborates a cue in some iterative manner until a specific and effective cue is generated, enters the retrieval process and knowledge in long-term memory is then accessed”. He added that “It is important to note that this ‘generative process’ does not happened [sic] consciously. Sometimes aspects of it may enter consciousness, sometimes it may be completely non-conscious” and “I could go on but the point is plain and that is that we cannot introspect on non-conscious memory retrieval process although we can be aware of some of their outputs”. Here we develop this statement: our participants may feel like they have initiated retrieval, but all that is known to them is what fragments may come to mind or how fluently the cue has been processed.

Additional information

Funding

This work was supported by a studentship funded by the Agence Nationale de la Recherche [ANR-15-IDEX-02 METASTORY] project awarded to Céline Souchay (Université Grenoble Alpes), Claire Barnes and Andrea Tales (Swansea University).