Did you ever happen to talk about a conversation you had with a friend just to realize that you cannot quite remember all that was said? Our memory of conversations is not always perfect. In fact, we tend to remember only the gist of what was said, rarely with word-by-word memory, and we often remember our own thoughts and inferences during the conversation as if we had said them. This is because a lot is going on during a conversation: one speaks, listens, looks at the gestures and expressions of the conversation partner, thinks and makes inferences about what the partner means, etc. It’s hardly possible to remember everything accurately. 

How well someone remembers an event, or, in this context, a conversation, is tested in experiments by asking participants to freely recall everything mentioned. This method leads to quite accurately remembered information because you tend to report only what you remember. However, the amount of remembered information is surprisingly low (it can decrease down to 6% after a few days). Another way of testing memory for conversations involves cueing participants by asking them about a specific piece of information that was mentioned or who said something. This approach has the advantage of uncovering memory for items that are not spontaneously recalled during the free recall test but can also introduce false memories. For example, one may then be cued to mention the inferences they had about the interlocutor’s intentions as if they were actually mentioned information. Similarly, we may tend to report something we expected the interlocutor to say (but never actually said). This is especially relevant in the context of witness statements, where witnesses may misremember the exact wording of conversations and eventually report their own impressions rather than give an unbiased perspective. The implications of this become important in the context of reported conversations with ex-presidents that are investigated, such as with Nixon during the Watergate investigation or, more recently, with Trump. In the Nixon case, when tapes of the conversation surfaced, it turned out that many of the reported details were wrong, although fundamentally, what was reported had indeed happened.

A caricature by Jack Davis captures the Watergate blame game.
Having clear and sound depositions is crucial during investigations. In Davis’s Time cover, the conspirators encircling Nixon are, clockwise from upper left: James McCord, Jeb Magruder, H.R. Haldeman, John Dean, John Mitchell, and Maurice Stans. (National Portrait Gallery, Smithsonian Institution; gift of Time magazine; Estate of Jack Davis 1973)

McKinley, Brown-Schmidt and Benjamin (2017) tested how well participants remember what was said in conversation, who said what, and to whom something was said. They found that memory for the content of the conversation, such as the objects that were mentioned, was better for speakers than listeners. In other words, you are more likely to remember what you said than what you heard in a conversation. Other research shows that people tend to remember information in focus better, such as answers to questions they posed. For example, you may be more likely to remember the answer to a question you asked than a question that was asked to you. McKinley and colleagues also found that the more the speaker and the listener converge on mutual knowledge and understanding (developing what is usually called common ground), the better their memory. The memory for the context of the conversation, such as who told you something or to whom you spoke, is usually worse than the memory for the content of the conversation. This can lead to cases where you remember an interesting piece of information but don’t remember who told you. With the source of information gone, one may also forget how reliable this statement was – which could be an interesting mechanism potentiating the spread of fake news.

Therefore, memory for conversation is fallible due to the many things we pay attention to when we interact. This can have consequences for the reliability of witness statements during trials in legal contexts and when adjudicating who came up with an idea when assigning awards such as the Nobel Prize. Overall, we can be confident about the gist of what we talked about, but we may rely too strongly on our impressions for the details of the conversation.  

Further reading:

Brown-Schmidt, S., & Benjamin, A. S. (2018). How We Remember Conversation: Implications in Legal Settings. Policy Insights from the Behavioral and Brain Sciences, 5(2), 187–194. https://doi.org/10.1177/2372732218786975

McKinley, G. L., Brown-Schmidt, S., & Benjamin, A. S. (2017). Memory for conversation and the development of common ground. Memory & Cognition, 45(8), 1281–1294. https://doi.org/10.3758/s13421-017-0730-3

Zormpa, E. (2020). Memory for speaking and listening. PhD Thesis. Radboud University Nijmegen, Nijmegen, The Netherlands. 


Cover image: A conversation between Roberto Benigni (left) and Steven Wright (right) from the movie Coffee and Cigarettes by Jim Jarmusch, United Artists, 2004.