Abstract

Much of a meeting’s value lies in what happens after it, yet what was said, who said it and who agreed to do what fade quickly. This article reviews four bodies of evidence. First, meeting interaction is associated with team and organizational outcomes, while people recall only a small part of a conversation and groups remembering together recall less than their members would separately (Kauffeld & Lehmann-Willenbrock, 2012; Stafford & Daly, 1984; Weldon & Bellinger, 1997). Second, memory for who said what is a reconstruction that errs: people take others’ ideas as their own and credit themselves with more of a joint product than others do (Johnson et al., 1993; Brown & Murphy, 1989; Ross & Sicoly, 1979). Third, effort drops in collective work, and evaluation potential is among the strongest moderators of that drop (Karau & Williams, 1993). Fourth, notes help partly through later review, and a popular claim about how notes should be taken did not hold in a direct replication (Kiewra, 1985; Kobayashi, 2006; Morehead et al., 2019). On this evidence, Mars turns a meeting recording into minutes and an editable report deck, puts every action item in a table of owner, task, due date and recording time, and gives each speaker a column of points with their times. We also state the limits of this evidence: it supports the design direction and is not an experimental test of Mars itself.

Keywords: meeting science · conversational memory · collaborative inhibition · source monitoring · egocentric bias · social loafing · note-taking and review · AI meeting notes

1. Introduction

Meetings are a large part of organizational life, and their quality is not trivial. In 92 videotaped regular team meetings, teams that showed more functional interaction, such as problem-solving and action planning, were more satisfied with their meetings; better meetings were associated with higher team productivity, and constructive interaction was related to organizational success 2.5 years later (Kauffeld & Lehmann-Willenbrock, 2012). In two surveys of full-time employees (N = 676 and N = 304), perceived meeting effectiveness had a strong, direct relationship with job attitudes and well-being (Rogelberg et al., 2006). Design matters too: across two further studies (N = 958 and N = 292), agenda use predicted perceived effectiveness, with attendee involvement as a key mediator (Leach et al., 2009).

A meeting is only as useful as what people carry out of it. This article asks why the content of a meeting, its decisions, its commitments and who said what, is so easily lost or misremembered, and how those answers translate into the design of a meeting assistant.

Method: this is a narrative literature review. We favoured peer-reviewed work accessible through a DOI, and preferred meta-analyses, multi-study designs and direct replications. Every factual statement carries its source, and the full reference list with original links follows the article.

2. We remember little of what was said

Conversations are hard to remember. In a study of memory for natural conversations, participants recalled only about 10% of what was said after just five minutes; recall mode and memory goals affected how much they recalled, and they remembered more of their partner’s comments than their own (Stafford & Daly, 1984).

Remembering together does not simply add memories up. In two experiments, groups recalling collaboratively recalled more than individuals but less than the pooled recall of the same number of individuals working alone, a pattern called collaborative inhibition (Weldon & Bellinger, 1997). A review of this research adds that collaboration can lead to forgetting and increase memory errors, but can also improve memory under proper conditions (Rajaram & Pereira-Pasarin, 2010).

Discussion can also fix the wrong picture in place. In a simulated political caucus, discussion tended to be dominated by information members already shared and by information that supported their existing preferences; members’ recall after discussion showed that it perpetuated, rather than corrected, their distorted views of the candidates (Stasser & Titus, 1985).

Based on this evidence, Mars does not ask people to reconstruct a meeting from memory. Attach the recording and say “turn this recording into a report deck”: Mars listens to the whole meeting first, works out the agenda, decisions, action items and each speaker’s views, and lays them out as an editable deck of meeting summary, key decisions, action items, each side’s views, risks and open questions, and the next step (Stafford & Daly, 1984; Weldon & Bellinger, 1997).

3. “Who said that?” is easy to get wrong

Remembering where a memory came from is a judgment of its own. The source monitoring framework holds that people attribute a memory to a source from the qualities of the experience, and that these judgments are subject to error and disruption; cryptomnesia and misattributed familiarity are among the phenomena it explains (Johnson et al., 1993).

Group settings make such errors concrete. In three experiments in which participants generated category examples in turn with others, they later reproduced others’ responses as their own despite instructions to avoid it, and the most likely source of a plagiarized response was the person who had responded just before them (Brown & Murphy, 1989). Observers of small-group discussions tended to encode people by sex and race and to minimize differences within those categories, with at least partial support for these hypotheses across three studies (Taylor et al., 1978).

Credit is also remembered unevenly. In five experiments, among them married couples and discussion groups, people recalled their own contributions to a joint product more readily and took more responsibility for it than others attributed to them (Ross & Sicoly, 1979).

Based on this evidence, Mars ties each point to a voice and a moment. Its own speaker recognition tells the voices in a recording apart: ask for minutes and you first get a who-said-what transcript table of speaker, time and words, where a tap on the time jumps the recording to that second; ask for a deck and the views slide gives each speaker a column, every point with its time. Names come only from a name spoken in the recording, a participant you named or a mapping you confirmed; otherwise they are Speaker 1 and Speaker 2 (Johnson et al., 1993; Brown & Murphy, 1989).

4. A commitment without an owner fades

Shared work invites diffused effort. A meta-analysis of 78 studies found social loafing, the tendency to expend less effort when working collectively than individually, to be robust across tasks and populations; evaluation potential, expectations of co-worker performance, task meaningfulness and culture had especially strong influence on it (Karau & Williams, 1993).

Within meetings, action planning is part of the functional interaction associated with meeting satisfaction and team productivity (Kauffeld & Lehmann-Willenbrock, 2012). The effect of being answerable is not uniform, however: a review of accountability research is organized around the conditions under which accountability attenuates, has no effect on or amplifies cognitive biases (Lerner & Tetlock, 1999).

Together, the evidence suggests that “I’ll take care of it”, said in passing, is easily lost unless it is attached to a named person and a recorded moment.

Based on this evidence, Mars puts every action item in a table of owner, task, due date and time in the recording, one item per row. A promise made in passing and a standing duty count too, key decisions keep any “if this, then that” condition whole, and before building the deck Mars goes back over the meeting for commitments and decisions only and adds any it missed (Karau & Williams, 1993; Kauffeld & Lehmann-Willenbrock, 2012).

5. Notes help when they are reviewed

Note-taking research distinguishes two functions. Its conclusions are that taking notes is beneficial independent of review, the process function, and that reviewing them is additionally beneficial, the product function (Kiewra, 1985). A meta-analysis of 33 studies found the combined effect of taking and reviewing notes to be substantial; providing a framework or instructor’s notes enhanced it more than pre-training or verbal instruction alone (Kobayashi, 2006).

How notes should be taken is less settled than often claimed. Three studies reported that students who took notes on laptops did worse on conceptual questions than those who wrote longhand, attributing this to verbatim transcription (Mueller & Oppenheimer, 2014). A direct replication and extension found that performance did not consistently differ between groups, including a group that took no notes; differences shrank further after students studied their notes, a meta-analysis of the direct replications showed small, nonsignificant effects favouring longhand, and the authors concluded that naming a superior method seems premature (Morehead et al., 2019).

Based on this evidence, Mars treats minutes as something to check and come back to. Every row carries its [mm:ss] time, the speaker notes say who said it and what was said, and a tap on a time jumps the recording to that moment, so each point can be checked against the recording (Kiewra, 1985; Kobayashi, 2006).

6. From evidence to design

These findings point to one conclusion: a tool that helps people keep what a meeting produced must hold the whole conversation, keep each point attached to its speaker, give each commitment an owner and a date, and make the record easy to check. Mars meets each requirement as follows:

1. Little of a conversation is recalled, and groups recall less together → minutes and a report deck from the recordingAttach a meeting recording and ask for a report deck: Mars listens to the whole meeting first, then builds an editable PowerPoint deck of meeting summary, key decisions, action items, each side’s views, risks and open questions, with a closing slide for the next step or next meeting. A section with nothing to say is left out rather than padded (Stafford & Daly, 1984; Weldon & Bellinger, 1997; Stasser & Titus, 1985).
2. Memory for sources errs → speaker recognition with a column per speakerMars’s own speaker recognition tells the voices apart, and the views slide gives each speaker a column, every point with its recording time. Names come only from the recording, from you or from a mapping you confirmed, and with only one voice Mars never splits the meeting into speakers. Once you confirm which contact a voice is, their lines and the follow-ups land on that contact (Johnson et al., 1993; Brown & Murphy, 1989; Ross & Sicoly, 1979).
3. Collective work diffuses effort → an owner and a due date for every action itemThe action items slide lists owner, task, due date and time in the recording, one item per row; a long list continues on a second slide instead of being merged. Promises made in passing and standing duties such as a weekly sync count as action items, and decisions keep their conditions (Karau & Williams, 1993; Kauffeld & Lehmann-Willenbrock, 2012).
4. Notes help through review → every row traced to the recordingEvery row carries its [mm:ss] time and the speaker notes say who said it, so you can jump straight to that moment and check it. You download the deck and decide who gets it; Mars never sends it to anyone (Kiewra, 1985; Kobayashi, 2006).

7. Limitations

First, the studies cited here test general regularities of memory and group work, not Mars itself. They support our design direction but do not prove that using Mars produces the same effects; the product’s effect needs its own evaluation.

Second, several findings come from laboratory tasks and student samples, such as category generation in turn (Brown & Murphy, 1989), recall of pictures, words and a story (Weldon & Bellinger, 1997), a simulated caucus (Stasser & Titus, 1985) and lecture note-taking (Kobayashi, 2006; Mueller & Oppenheimer, 2014; Morehead et al., 2019), so generalizing to workplace meetings calls for care. The field studies of meetings are correlational and do not by themselves establish causation (Kauffeld & Lehmann-Willenbrock, 2012; Rogelberg et al., 2006; Leach et al., 2009).

Third, popular claims about notes are often stated more firmly than the evidence allows: the longhand advantage was not consistently reproduced in a direct replication (Morehead et al., 2019), and accountability does not reduce bias under every condition (Lerner & Tetlock, 1999). We cite these findings as directions, not as fixed rules.

Fourth, an automatic record can itself misattribute. Mars therefore writes names only from the recording, from a participant you named or from a mapping you confirmed, and otherwise writes Speaker 1 and Speaker 2; every point keeps its recording time so it can be checked (Johnson et al., 1993).

8. Conclusion

Meetings are rarely forgotten all at once; they fade a detail at a time, while “who said that” and “who owns this” drift. The research shows that limited memory for conversations, errors in remembering sources and diffused responsibility in groups together drive this loss, and that a complete record, points tied to speakers, named owners and a record that can be reviewed are the matching remedies. Mars meeting minutes, report decks, action item tables and speaker views are designed on this evidence.

References

Listed in APA 7th edition style, each with its original DOI link.

1. Brown, A. S., & Murphy, D. R. (1989). Cryptomnesia: Delineating inadvertent plagiarism. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15(3), 432–442. https://doi.org/10.1037/0278-7393.15.3.432
2. Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114(1), 3–28. https://doi.org/10.1037/0033-2909.114.1.3
3. Karau, S. J., & Williams, K. D. (1993). Social loafing: A meta-analytic review and theoretical integration. Journal of Personality and Social Psychology, 65(4), 681–706. https://doi.org/10.1037/0022-3514.65.4.681
4. Kauffeld, S., & Lehmann-Willenbrock, N. (2012). Meetings matter: Effects of team meetings on team and organizational success. Small Group Research, 43(2), 130–158. https://doi.org/10.1177/1046496411429599
5. Kiewra, K. A. (1985). Investigating notetaking and review: A depth of processing alternative. Educational Psychologist, 20(1), 23–32. https://doi.org/10.1207/s15326985ep2001_4
6. Kobayashi, K. (2006). Combined effects of note-taking/-reviewing on learning and the enhancement through interventions: A meta-analytic review. Educational Psychology, 26(3), 459–477. https://doi.org/10.1080/01443410500342070
7. Leach, D. J., Rogelberg, S. G., Warr, P. B., & Burnfield, J. L. (2009). Perceived meeting effectiveness: The role of design characteristics. Journal of Business and Psychology, 24(1), 65–76. https://doi.org/10.1007/s10869-009-9092-6
8. Lerner, J. S., & Tetlock, P. E. (1999). Accounting for the effects of accountability. Psychological Bulletin, 125(2), 255–275. https://doi.org/10.1037/0033-2909.125.2.255
9. Morehead, K., Dunlosky, J., & Rawson, K. A. (2019). How much mightier is the pen than the keyboard for note-taking? A replication and extension of Mueller and Oppenheimer (2014). Educational Psychology Review, 31(3), 753–780. https://doi.org/10.1007/s10648-019-09468-2
10. Mueller, P. A., & Oppenheimer, D. M. (2014). The pen is mightier than the keyboard: Advantages of longhand over laptop note taking. Psychological Science, 25(6), 1159–1168. https://doi.org/10.1177/0956797614524581
11. Rajaram, S., & Pereira-Pasarin, L. P. (2010). Collaborative memory: Cognitive research and theory. Perspectives on Psychological Science, 5(6), 649–663. https://doi.org/10.1177/1745691610388763
12. Rogelberg, S. G., Leach, D. J., Warr, P. B., & Burnfield, J. L. (2006). “Not another meeting!” Are meeting time demands related to employee well-being?. Journal of Applied Psychology, 91(1), 83–96. https://doi.org/10.1037/0021-9010.91.1.83
13. Ross, M., & Sicoly, F. (1979). Egocentric biases in availability and attribution. Journal of Personality and Social Psychology, 37(3), 322–336. https://doi.org/10.1037/0022-3514.37.3.322
14. Stafford, L., & Daly, J. A. (1984). Conversational memory: The effects of recall mode and memory expectancies on remembrances of natural conversations. Human Communication Research, 10(3), 379–402. https://doi.org/10.1111/j.1468-2958.1984.tb00024.x
15. Stasser, G., & Titus, W. (1985). Pooling of unshared information in group decision making: Biased information sampling during discussion. Journal of Personality and Social Psychology, 48(6), 1467–1478. https://doi.org/10.1037/0022-3514.48.6.1467
16. Taylor, S. E., Fiske, S. T., Etcoff, N. L., & Ruderman, A. J. (1978). Categorical and contextual bases of person memory and stereotyping. Journal of Personality and Social Psychology, 36(7), 778–793. https://doi.org/10.1037/0022-3514.36.7.778
17. Weldon, M. S., & Bellinger, K. D. (1997). Collective memory: Collaborative and individual processes in remembering. Journal of Experimental Psychology: Learning, Memory, and Cognition, 23(5), 1160–1175. https://doi.org/10.1037/0278-7393.23.5.1160

Cite this article

Mars Research. (2026). Why meetings are forgotten: minutes, action items and who said what. MarsDataAI. https://marsdata.ai/research/meetings-memory-and-commitments