The effects of context on memory and metacognition
Whenever we learn something new, try to recall something, or want to report what we remember, this action takes place in the presence of sounds, sights our thoughts, sometimes also other people. Everything that co-occurs with the (meta)cognitive tasks we are completing but is not directly related to them can be thought of as the context in which those tasks are performed.
In memory tasks, context often becomes encoded together with the to-be-remembered information. We have shown how remembering context shapes predictions of future memory performance: when revising, people believe they will be able to retrieve later more information if it is accompanied at restudy by the same context in which it was first encountered (so-called 'reinstated context'). Reinstating context at test can also make us more confident in the answers we provide. However, context can also mislead us sometimes, resulting in less adaptive memory and metacognitive strategies. Context reinstatement can be dangerous when the queried information is, in fact, not known: it might make us more willing to guess the answer rather than say - correctly - 'don't know' or 'don't remember'. Also, familiar context - one that is linked to a different event than the one we are trying to remember - often leads to metacognitive illusions of knowing more than we really do, as revealed in inflated predictions of future recognisability of information we currently cannot recall and confidence in the correctness of one's answer, as well as lower rates of 'don't know' responses. Finally, we have demonstrated that not only does context information influence memory for items encoded in that context, but that the reverse effect also occurs: contexts are remembered better after being reinstated at test rather than re-presented with other familiar information.
Interpretations of metacognitive judgements
Metacognitive judgements are commonly elicited in experimental procedures: for example, a participant in a psychology experiment might be asked to rate their confidence that their answer is correct (a retrospective confidence judgement, RCJ), to predict their future memory for some information (a judgement of learning, JOL), or to try to assess whether they would be able to later recognise an item that can't be recalled (a feeling-of-knowing judgement, FOKJ). These ratings are most often provided on scales, such as 1-to-n or 0-100%.
Our research has looked at the bases and interpretations of metacognitive judgements. For RCJs, we have looked at what determines confidence in tasks in which people are asked to choose a correct answer from among the answer and a lure (a so-called two-alternatives forced-choiced recognition test, or simply 2AFC). It was commonly believed that confidence is a relative judgement based on the difference in the strength of evidence between the two available options: if one answer has much stronger evidence than the other, then RCJs should be high; if both answers have comparable evidence, then RCJs should be low. We have shown, however, this is not the case. Instead, confidence in 2AFC generally seems to be an absolute rating reflecting the strength of evidence for the chosen answer only.
We have also investigated the meaning of JOLs. In a series of studies, we have shown that, contrary to the commonly held view, JOLs are not assessments of probability of future retrieval: instead, they are based on how studied items are rank-ordered in terms of their evidence for future recall and are best thought of as ordinal ratings of confidence. This casts doubt on some measures often used in metacognitive research, such as those of over- or underconfidence, which require an assumption that ratings that feed into the calculations are made on an interval or ratio scale. We have also introduced binary JOL-type ratings such as yes/no JOLs, betting tasks, or 0/100% JOLs (see here and here) that can be used to gain additional insight into people's assessments of their future memory performance.
Social influence on memory and metacognition
In everyday life, we share experiences with other people. As a result, we often gain new information and later retrieve it from memory in the presence of others. For example, we might attend one friend's birthday party (thus encoding the whole experience), and then tell another friend what happened at the party (retrieving information from memory). Other people can also contribute to this retelling of a past experience, adding information forgotten by us or never encoded in the first place. These contributions from others can affect what and how we remember and what we later report when asked a memory question. An issue of practical importance is what determines the willingness to incorporate information provided by others into our own memory reports.
In our research, we have looked at how the overall reliability of an external source - presented as another person - affects memory and metacognitive decisions made in a recognition memory test. We have shown that, in general, information from external sources influences people's memory decisions regardless of whether the source's memory is reliable (almost always accurate) or not (the responses are random). This effect is independent of how good people's memory performance is: the same patterns of responding can be found in memory tasks that vastly differ in the amount of evidence supporting recognition decisions (manuscript under review). However, when participants get feedback on the correctness of their own responses, they continue to trust the accurate source, but discount information from the random one. This pattern can be found both in their recognition responses and metacognitive decisions whether to report or withhold the chosen answer, resulting in metacognitive resistance. Another line of our research has demonstrated that the effect of source correctness is moderated by the emotional expressions of the sources: people are more inclined to rely on the memory of a person who seems happy rather than annoyed, and for some this pattern holds even when the happy person provides responses of no informational value (manuscript in preparation).
Memory under distraction
Noise is a ubiquitous feature of modern life, as anyone who has tried at least once to read on a train could testify to. The presence of noise is not without consequences for our memory functioning. Decades of research have shown that noise, particularly in the form of human speech, impedes learning. This seems to occur because the human cognitive system routinely screens the environment for important information, causing attention to be at least partly allocated towards distraction and away from focal tasks such as learning.
The investigations of the effects of auditory distraction on memory generally follow two paths. In the first strand of research, the specific mechanisms involved in processing distraction are elucidated. For example, it has been argued that auditory distraction triggers inhibitory mechanisms directed towards semantic contents of distraction, which exert their lasting influence also when currently to-be-ignored information becomes subsequently necessary for performing a cognitive task. However, in our investigation we have questioned the idea assigning inhibition such a role in dealing with auditory distraction. Instead, we have promoted an idea that processing auditory distraction disrupts higher-order processes such the creation and execution of an effective retrieval strategy. In accordance with this hypothesis, we have shown that there are circumstances when auditory distraction augments memory: these are situations in which auditory distraction is tailored in such a way as to support rather than disrupt an effective retrieval plan.
The second strand of research on auditory distraction is generally devoted to ways by which the negative effects of distraction can be mitigated. Here we looked at metacognitive strategies people may deploy to counteract the memory impairment brought about by noise accompanying learning and/or retrieval. We have looked at decisions whether candidate responses should be volunteered in a memory report but we found that people are generally unable to use response withholding to improve memory performance under distraction. We have also examined whether people would compensate for noise by extending study times, but again we found no such strategy and indeed we found evidence that study times are curtailed under distraction. The quest for effective strategies for dealing with noisy environments in which learning takes place is thus ongoing.