Group Study and Collaborative Inhibition: Two Heads Recall Less Than One

Introduction

It is the night before the exam and four of you are around a kitchen table. Someone reads a question aloud. Someone else answers before you have finished thinking. Names, dates and definitions fly back and forth, and after two hours it feels like the best revision session of the term.

Here is the uncomfortable part. If the same four people had spent the first twenty minutes recalling alone, on paper, and only then compared notes, the table as a whole would very likely have produced more.

Psychologists call the shortfall collaborative inhibition. Put bluntly, two heads recall less than one.

Read literally, that line is false, so let us fix it now. A group almost always remembers more than any single person in it. Two heads together recall less than the same two heads recalling separately and then pooling what they found.

That correction changes the question. Why does talking at the same table cost memories, and can you keep what the group gives you without paying that price?

You can, and the answer comes down to the order you do things in. Getting there takes us through seventy years of research, some pilots and some married couples.

The Arithmetic Everyone Gets Wrong

The comparison at the heart of this field is subtle, and most popular write-ups skip it.

Imagine you and two friends study the same list of 30 words. Alone, each of you recalls about half. Because you remember different words, your three solo lists might together contain 24 different words, counting each word once. Researchers call that pooled total a nominal group. It is a group on paper only, and its members never spoke.

Now put three similar people in a room and let them recall the list together, out loud. They will usually beat what any one of them could manage alone. They will usually fall short of the pooled 24. The gap between the real group and the paper group is collaborative inhibition.

Mary Susan Weldon and Krystal Bellinger gave the effect its name in 1997 [1]. In their first experiment small groups recalled pictures and words, and in their second they recalled Bartlett's old folk tale The War of the Ghosts. Both times the collaborating groups recalled more than individuals but less than nominal groups, and in the story task they leaned harder on their single best member.

Keep that pattern in view. Better than one person, worse than the same people apart.

Yuker had shown the first half of that pattern, a group outscoring an individual, as far back as 1955 [2]. Across four experiments published in 1992, Meudell and colleagues then found that two people recalling together beat one person only by pooling what each of them already held, and that the conversation never produced a memory neither partner had [3].

Collaboration added memories up. It did not create new ones. Once researchers stopped comparing a group with a person and started comparing it with its own members working apart, that modest advantage turned into a deficit.

How Big the Shortfall Is

A strange result that appears once is a curiosity. This one appears almost everywhere anyone has looked for it.

In 2016 Stéphanie Marion and Craig Thorley pooled 75 effect sizes from 64 studies and found the deficit reliable [4], confirming what pair studies had shown since 1995, when Andersson and Rönnberg found that pairs never reached their predicted potential, though friends and more complex tasks shrank the loss [5]. In the meta-analysis the deficit was larger in bigger groups, with word lists that had no categories, when members talked freely in any order, and when they were strangers.

Each of those conditions makes it harder to keep your own way of searching memory. Hold on to that clue.

A larger synthesis followed. In 2024 Luo and colleagues gathered 140 independent studies with 185 effect sizes and 19,492 people, published between 1996 and 2023 [6]. The pooled effect size was 0.54, falling to 0.40 after a standard correction for studies that may be missing from the record, and it shifted with the gender mix of a group, whether members knew each other, culture and how well people monitored their own memory [6].

That is a moderate effect, not a catastrophe. You will not forget half your notes because a friend is in the room. But it is consistent, and it runs against what almost everyone expects.

Seventy Years of Asking Whether Two Heads Are Better

The finding arrived in pieces, from memory labs, social psychologists and brainstorming studies running in parallel for decades.

Notice how late the classroom arrives on that list, which matters if you plan to use any of this for your own revision. For most of its history this has been a laboratory finding about word lists.

By 2010 the field's framework review argued that collaboration can also improve memory under the right conditions, and set out the mechanisms behind both costs and benefits [7]. A 2008 review had asked which features of groups, and of memories, push people towards a shared version of the past [8].

The Experiment That Pointed at Retrieval

Why would a group lose items that its members, working alone, would have found? The first serious answer came in the same year as the name.

When you recall a list alone, you settle into a route through it. Maybe you go category by category, fruit then tools then animals. Maybe you start with the first words you studied and work forward. Every time somebody calls out an item that does not fit your route, you are pushed off it.

Basden and colleagues tested that idea in four experiments published in 1997, using categorised word lists and groups that took turns to recall [9]. The deficit appeared when the categories were large and when category names were handed out as cues. When each member was made responsible for a different part of the list, or when everyone had to recall category by category, the real groups recalled as much as the nominal ones.

Take away the competing routes. The deficit goes with them. That was the first strong sign that the problem lies in retrieval, not in learning.

Finlay, Hitch and Meudell pushed the idea further in 2000, working with pairs [10]. Items a pair failed to produce together came back when the same people recalled alone afterwards, so the memories had not been lost, only blocked for a while. Giving every item its own cue removed the deficit, and having both partners learn the list in the same order shrank it, because it gave them similar routes from the start.

If you have ever lost your train of thought because a friend finished your sentence the wrong way, you have felt this mechanism. It has a clumsy name, retrieval strategy disruption, and for two decades it has been the leading explanation.

It Is Not Laziness

The obvious alternative is motivation. In a group nobody carries the result alone, so perhaps people simply try less hard.

Social psychologists call that social loafing, and a 1993 meta-analysis of 78 studies by Karau and Williams found that people reliably put in less effort when their work is pooled with other people's [11]. Yet when Weldon, Blair and Huebsch tested loafing directly across five experiments in 2000, paying people for performance, changing the recall criterion, making each member accountable, building cohesion and varying the gender make-up of groups, extra motivation sometimes raised the overall amount recalled and never removed the deficit [12].

So effort is not the missing ingredient. You can care a great deal about the result and still lose items to the group.

A close cousin of the effect had turned up earlier in brainstorming research, where Diehl and Stroebe found across four experiments in 1987 that production blocking, waiting for your turn while your idea fades, accounted for most of the ideas real groups lost [13]. For memory, waiting is not the whole story. In a 2004 study by Wright and Klumpp, people who took turns but could not see each other's answers did not differ reliably from nominal groups, while people who could see the answers showed the usual deficit [14].

The authors concluded that the damage comes from the content of other people's recall. It is what you hear, not how long you wait, that derails you.

Coordinating one shared answer carries a separate cost. In a 2008 study in which three-person groups recalled a film clip, face-to-face groups more often dropped items at the group level, while groups linked by computer more often filtered themselves before speaking, and the face-to-face groups helped their members recall more afterwards [15].

That last detail is a first hint that the group session has another side.

Cracks in the Leading Explanation

Retrieval strategy disruption is still the best supported account, and a 2024 review of the empirical factors described a consensus that cognitive processes matter more than social ones [16]. The sharpest challenge to it came from Barber, Harris and Rajaram in 2015, in an experiment where 108 participants studied words in groups of three but each member studied different words, so nothing was shared to compete over [17]. Collaboration still cost them, and the loss carried over to a later solo recall test and, in a second experiment, to a recognition test where people only had to pick out words they had seen.

Pure disruption predicts those items should bounce back. They bounced back only partly. Barber's team suggested that recalling some items actively suppresses related ones, the same suppression that produces retrieval-induced forgetting, so hearing a partner recall may do the same to the words you were about to say.

Two newer results point elsewhere. In 2021 Ekeocha shielded the members of three-person groups from each other's contributions, and the deficit stayed [18]. In a 2024 study of 320 undergraduates in Italy and Spain, Conte and colleagues found the deficit even in collaborative groups whose recall was organised as neatly as individuals' recall [19].

If hearing other people were the whole cause, shielding should have helped. The 2024 team suggested divided attention as the missing piece: in a group you track other people while searching your own memory, and both draw on the same limited attention.

Some of the loss looks like caution. In 2013 Hyman, Cardwell and Roy found that collaborating pairs worked harder to avoid errors, which trimmed their correct recall too, and that they recalled fewer words early in a session but more later on [20].

Part of the deficit, then, is a choice about caution rather than a failure of memory. A group that would rather be right than complete will say less.

Computational models now reproduce the effect with no motivational assumption at all. A 2024 model treats memory search as a walk through a mental context and shows collaborators' contexts drifting together until they search the same ground, a convergence linked to lower recall [21]. Brain recordings fit the same picture, with a 2022 study finding electrical signatures of interference early in collaborative recall [22] and a 2023 study finding partners' prefrontal activity falling into step in a way that predicted how similar their memories became [23].

None of this requires anyone to be lazy or distracted on purpose. The loss falls out of how memory search behaves when several searches share one conversation.

So disruption explains a great deal, and probably not everything. The 2015, 2021 and 2024 results each found something it does not cover, and the field has not agreed on what fills the gap.

When the Gap Widens and When It Closes

If disruption is the main engine, anything that keeps people on their own routes should shrink the gap, and anything that mixes routes should widen it. Mostly, that is what happens.

Group size is the cleanest live dispute. The 2016 meta-analysis found that bigger groups lose more [4], but Gates, Suchow and Griffiths recruited 1,138 people online for their first experiment, placed them in chat groups of 2, 3, 4, 8 and 16, and found no sign that the deficit grew with size despite plenty of statistical power [24].

Online chat may change how groups recall, or the older lab studies may have been too small. Nobody has settled it. Be wary of anyone who says they have.

The task matters too, and the most recent result carries a twist. In 2023 Saraiva, Albuquerque and Garrido asked pairs to recall a word list either in any order or in the exact order it was learned, which is the ordered recall examined in research on the serial position effect.

With 38 pairs on each task in the first experiment, pooled pairs recalled 66 percent of the words in free recall against 51 percent for real pairs, and in serial recall the figures were 50 and 43 percent [25].

The gap shrank from 15 points to 7. It is worth seeing why. Real pairs lost 8 points when order was required, while pooled individuals lost 16. Ordered recall did not make the groups better. It made people working alone worse, by taking away their freedom to follow their own route.

In the second experiment the solo recallers were also made to take turns, and the gap fell to 4 points in free recall, 60 against 56 percent, and to 3 points in serial recall, where it was no longer statistically reliable [25].

People differ, too. In a 2023 study of 120 people, Hood and colleagues measured working memory capacity before the recall task and found that people with lower capacity lost more in collaboration and showed no benefit afterwards, while people with higher capacity gained from having collaborated, and collaboration cut errors for both [26].

So working memory decides part of who gains from a group. The same session that helps one member can leave another a little worse off.

How the material was learned can remove the gap altogether. In two experiments published in 2013, groups that studied the material together before recalling it showed no collaborative inhibition, whether they were strangers or friends, and their joint recall contained fewer intrusions [27]. Pairs in which one partner processed the words deeply and the other only superficially showed no deficit in a 2025 study, apparently because their strategies stopped colliding and each could cue the other [28].

What these conditions share is alignment. When you and your partners come to the material the same way, or in complementary ways that cue each other, your searches stop colliding.

Going online moves the baseline. In a 2022 online study with 48 participants per condition, Greeley, Peña and Rajaram found no deficit at all, because people recalling alone at home did worse than people usually do in a lab. In a second experiment with 45 participants per condition, clearer instructions and an experimenter present brought solo recall back up, and the deficit returned [29]. Video calls with full sound and picture matched face-to-face collaboration in a 2024 comparison [30], and videoconference groups of three showed the usual deficit both straight away and after 24 hours, while being less suggestible than people working alone [31].

The lesson there is about the comparison, not the group. A deficit can vanish because the people working alone got worse, which is what happened with serial recall.

The Pilots and the Married Couples

The most instructive cases are the groups that do better together, because they show what the others are missing.

Meade, Nokes and Morrow asked expert pilots, novice pilots and non-pilots in 2009 to recall flight scenarios alone or with a partner of the same expertise [32]. Non-experts were disrupted by collaboration, as usual, but experts benefited, and their transcripts showed why: they explicitly acknowledged what a partner had said by repeating it back, and then elaborated on it.

That is the disruption account turned inside out. Experts share a structure for the material, so a partner's words land on a route they were already walking.

Long relationships can produce the same thing. In 2011 Harris and colleagues interviewed 12 older married couples in depth and found that some showed the usual deficit while others recalled more together than apart, with shared strategy the best predictor of success [33]. A 2023 study of 30 pairs of couples and 30 pairs of strangers found reliable collaborative facilitation, a pair recalling more than its pooled members, only in the older, long-term couples [34], and when long-married couples recalled trips they had taken together, they produced fewer items than their potential but more specific episodic detail [35].

What counts as remembering well depends on what you are trying to remember. A couple reliving a holiday is not trying to maximise a list.

Age itself is not the problem. Older and younger adults showed the same pattern of costs and benefits in a 2011 comparison, including fewer false memories during group recall [36], and a 2023 systematic review of 29 studies found that age differences in collaborative learning shrank or disappeared when older learners had enough time or trials [37]. A 2026 review points out that structured reminiscence sessions boost recall in aged care [38].

The common thread is shared structure, built through expertise or decades of practice.

Decision research adds a related condition. Two people of similar ability who share how confident they are can detect faint visual targets better than the better of the two alone [39], and groups that follow their most confident member reach the right answer when most members already know it but the wrong one when few do [40].

Pooling pays off when partners are well matched and most of them already know the answer. A first session among near strangers cannot count on either, which is why its order of work matters.

The Other Side of the Ledger

So far this has been about loss. It is only half of the story, and the less useful half.

The same 2016 meta-analysis that confirmed the deficit ran a second analysis on what happens afterwards, across 27 effect sizes, and found that people who had recalled in a group usually remembered more on a later solo test, partly because they had been re-exposed to items they missed [4]. A 2024 meta-analysis of 64 studies, with 101 effect sizes and 13,398 participants from 11 countries, put that later benefit at a moderate size [41].

The group session lowers the group's total and raises each member's later memory. Both are true at once, and the order you do things in decides how much of each you get.

Order is where this becomes practical. In 2008 Blumen and Rajaram had people recall a list three times in different sequences, and recalling alone, then in a group, then alone again beat three solo recalls on the final test, as did two group recalls followed by a solo one [42].

Starting alone lets you lay down your own route before anyone can disturb it. Then the group fills your gaps without wrecking the route.

The benefit lasts. Blumen and Stern tested 60 younger and 60 older adults in groups of three, with two group recalls followed by two solo recalls, the last of them by phone 7 to 10 days later, and the advantage of having collaborated was still there in both age groups [43]. A 2024 study of 252 participants found the same persistence after 7 days for word lists and for prose passages, except that corrections of errors made during collaboration had faded by then [44].

The group helped people keep what was right, but it did not permanently cure what was wrong.

The most striking result in this line is about preparation. In 2011 Congleton and Rajaram had people learn a list either by restudying it or by repeatedly testing themselves, and repeated retrieval eliminated the collaborative deficit while letting people gain the most from hearing their partners' items afterwards [45].

Practising recall alone first, the habit behind the testing effect, appears to build the kind of organised memory that a group cannot knock over. Take one habit from this research, and make it that one.

There was even a benefit nobody went looking for. In a 2023 study with 96 participants, those who practised retrieval together reported less mind-wandering during learning than those who practised alone, alongside the usual deficit during practice and a benefit on the final test [46]. Structure has a cost, though, because when groups recalled under strict turn-taking in another 2023 study, the group still caught errors but the later memory benefit did not appear [47].

Groups Catch Errors, and Groups Spread Them

Memory errors travel through group recall in both directions, and this is where a study group can help you most or hurt you most.

In a 2008 study, when older and younger adults recalled household scenes with their spouses, partners caught and discarded each other's false memories during the conversation, and older couples held wrong answers back before saying them [48]. Groups recalling lists designed to provoke false memories produced more correct words than individuals without producing more false ones [49], and in a 2014 study where 57 participants recalled an emotional film in groups of three and 54 recalled it alone, the groups produced fewer details but also fewer errors [50].

Groups say less, and more of what they say is right, a trade that shows up with word lists, films and family memories alike.

The harmful direction is just as well documented. In 2001 Roediger, Meade and Bergman paired 24 undergraduates with a confederate who slipped occasional wrong items into a shared recall of scenes, and those planted errors turned up later in the participants' own solo recall [51].

The researchers called it the social contagion of memory, and it is one of the routes by which false memories form. A confident partner is the delivery system.

Who you recall with changes the risk. Romantic partners were more likely than strangers to adopt the other person's version of an event in a 2008 study [52], and friends in three-person groups showed more contagion than strangers and seemed to take each other's memories on trust, although the collaborative deficit itself vanished on a recognition test for friends and strangers alike [53].

Closeness lowers your guard. The closer you are to someone, the less carefully you check what they tell you.

Bigger groups and more pressure to produce items both raised false recall in a 2007 study of individuals, pairs, trios and quartets, and those false memories survived into a later solo test [54]. In a 2021 study, strict turn-taking produced more false memories than free conversation, and former group members carried those errors into later solo recall as shared false memories [55].

Rules about who speaks when are not a reliable error filter, since turn-taking helped in some studies and hurt in others. Checking against the source is the safer habit.

People also lose track of whose memory is whose. After collaborating, participants in a 2021 study claimed partners' memories as their own and assumed their own memories were shared more often than they were, biases absent when the same information came from a non-social source [56]. Silence does damage too, because in a 2007 study a listener later forgot related details that a selective speaker had left out, in paired word lists, stories and free conversation alike [57].

If you have ever argued with a friend about who actually said something first, you have seen the first of those biases at work. The silence effect means a partner who skips a topic can make it harder for you to recall it later.

Both forces run at once. In 2023, groups of three later showed both a memory boost for material they had studied and social contagion for material they had not, on the same final test, and both pulled members' memories closer together [58]. Turn-taking can still help with one kind of error, since in three experiments published in 2024 groups that took turns were less swayed by misinformation that had already been retracted [59].

For your study group the lesson is blunt. A confident wrong answer said out loud can become your answer, so when the group disagrees, the textbook decides and not the loudest voice.

How a Group Ends Up Remembering the Same Things

Recall together often enough and your memories start to match. Researchers call that overlap collective memory, and a 2012 review of conversational remembering traced how it builds up from what speakers say and leave out [60]. Retrieval is the stage that does most of the work: in a 2012 comparison, collaborating during study and during recall both lowered a group's output, the second more so, but only collaborating during recall built shared memories [61]. Even a partner's cues build overlap, since in a 2026 comparison cues from a partner produced more same-category answers, and more shared memory afterwards, than the same kind of cues presented as part of a list [62].

That sounds harmless. Usually it is. It is also how small errors become a group's agreed version of events.

The overlap reaches beyond content. Congleton and Rajaram showed in 2014 that collaboration reshapes the structure of what people remember, so former partners come to organise material more alike [63], and a 2025 study using representational similarity analysis found collaboration changing the global organisation of memory for both neutral and emotional material [64].

Your memory of a topic starts to take on the shape of the group's memory of it. That can help. It can also trap you.

There is a student-sized example. In 2019 Maswood, Rasmussen and Rajaram asked students to recall a real exam they had recently taken, and those who first recalled it together later remembered it with less negative emotion, a more positive tone and more overlap with each other [65].

Talking an exam over with friends may change how it feels in memory, not only what you remember of it.

Close groups can go further and divide memory between them, each person keeping certain things for the others, an idea Daniel Wegner proposed in 1987 as transactive memory [66]. It is what the most successful long-married couples seem to run on. A study group that meets all term may build a little of this, though nobody has tested it.

What Students Believe, and What They Do

Ask students and you hear something close to the opposite of the lab result. In a 2024 survey of 202 students, nearly half thought studying in a group was more effective, and more of them rated recalling in a group as better than recalling alone, yet most preferred to study alone and rarely used study groups [67]. When studying alone, 69.8 percent reread their notes and textbooks, while over 40 percent of those who studied in groups quizzed or taught each other [67].

So students are split on group study in general and optimistic about recalling together. That is precisely the activity the lab finds costly.

Researchers have noticed the gap between feeling and fact for a long time. Reviewing the field in 2011, Suparna Rajaram wrote that "despite subjective reports to the contrary, collaboration impairs a group's recall performance" compared with its potential, while members still improve afterwards [68].

Part of the reason is that hearing an answer feels like knowing it. A partner says the term, it sounds familiar, and you nod. Familiarity is not recall, and confusing the two is the trap described in the illusion of knowing.

The newest classroom data show what that trap can cost. In 2026 Williams and Denaro followed 502 of the 643 students enrolled in an introductory biology course and found that students chose encoding strategies such as rereading over retrieval strategies whether they studied alone or with friends, that heavier encoding went with lower exam scores, and that study group use, though popular, had no significant effect on exam scores [69].

The authors did not blame the friends. They concluded that studying together is not harmful but rests on an incomplete idea of what works, and suggested that study groups meet in weeks without exams, while students are still learning.

The Classroom Evidence Is Split

If you have sat a collaborative exam, you have met the classroom version. In the common two-stage format students sit a test alone and then immediately sit it again in small groups, which already has the order the lab recommends.

On the positive side, a randomised crossover study of 38 students in 2003 found better retention of material that had first been tested in groups [70]. In a large sociology course the group stage raised learning even with individual achievement held constant [71], and a 2014 crossover study in an earth and ocean science course found larger gains on later individual tests after group testing, with no difference between stronger and weaker students [72].

On that evidence the group stage looks like a gift. The other half of the evidence is less kind.

In a 2012 biology class split into groups of 92 and 104 students, group tests raised scores but made no difference on cumulative questions in later exams [73]. In a 2021 anatomy course with 196 students, two-stage testing did not improve retention either, though students seemed to gain teamwork and communication skills [74].

Courses differ in what they test and how long they wait, so a split result is no surprise. How the group stage is run looks especially important.

In a 2017 chiropractic programme, requiring students to justify every answer they changed during the group stage made collaborative testing more effective than simply letting them confer [75]. A 2021 study of two-stage exams found that when students compare their own first attempt with the group's discussion, they produce feedback for themselves that is richer than the comments their teachers wrote [76].

Both details turn the group stage into retrieval and comparison rather than passive listening. That is what the lab work would predict.

Wider evidence on small-group learning is more positive, but it measures something broader than recall. A 1999 meta-analysis of undergraduate science, mathematics, engineering and technology courses found that small-group learning raised achievement, improved attitudes and increased persistence [77], and a 2013 meta-analysis of 65 classroom studies from 1995 onwards reached the same verdict [78]. In a 2026 national study, study group participation went with higher knowledge scores, in a dose-dependent way, among 437 of the 471 medical students who took part [79].

That last result is observational, so it cannot show that the groups caused the gains. Keen students may simply join more groups.

The common belief that groups shine on harder material is contested too. In a 2026 online experiment with 152 participants, collaboration helped on low-complexity tasks and hurt on high-complexity ones, a result the authors themselves called surprising [80].

The classroom picture, then, is less settled than the lab one. Structured group stages help in some courses and not others, and nobody has yet tested a study group run in the order the lab recommends.

A Study Group Built Around the Evidence

Put the findings together and a session designed around them looks different from the kitchen table, though not by much. Mostly it changes the order.

Before anyone talks, each person studies and then tests themselves on paper, with the notes closed. That solo recall is the step that protected later memory in the 2008 sequence experiment, and retrieval practice before collaborating removed the deficit outright in 2011. It is also plain good studying, the core of active recall.

With everyone's own attempt on paper, the group pools. People read out what they produced, the gaps get filled, and this is where re-exposure does its work.

Disagreements deserve a rule of their own. They are where a group can quietly damage your memory. When two answers clash, nobody wins by confidence or volume, and someone opens the book. That one habit blunts social contagion, the main way group study plants errors.

Explaining belongs in the middle as well. When you have to say why an answer is right, as in the chiropractic exams where students justified each change, listening turns into retrieval. It is the same reason that explaining a topic in plain words works when you study alone.

The session ends where it started, with each person recalling alone. That final pass tells you what you know, not what the table knew.

Small groups are safer than large ones, at least in the lab, and people who organise material in similar ways lose less than strangers do. Timing matters too. A group that meets while you are still learning, as the 2026 biology study recommended, has more to give than one that meets the night before the exam to reread notes.

What Is Settled and What Is Not

Some of this you can treat as settled. Groups recall more than any one member and less than the same people working apart, and that holds across well over a hundred studies. Extra motivation does not remove the deficit. Recalling together usually helps your later solo memory, and group recall both prunes errors and spreads them.

Other questions are open, and it is worth knowing who stands where. Whether bigger groups lose more divides the 2016 meta-analysis from the 2021 online study of groups of up to sixteen people. Whether disruption is the whole mechanism is challenged by the work on unshared words, shielded groups and divided attention. Whether collaborative exams improve long-term retention depends on which course you look at. Even studying together is unsettled: it removed the deficit in two 2013 experiments, but collaborating during study lowered a group's output in a 2012 comparison.

Almost all of the core evidence also comes from word lists, recalled by undergraduates, over minutes or days. Nobody has yet run the definitive experiment on a real study group across a real term, so the honest position is confident about the mechanism and modest about the effect on your grades.

Conclusion

Back to the kitchen table. The four of you were not wasting your evening. You were doing something that has a cost and a benefit, in the wrong order.

The cost is real and well measured. Each time somebody calls out an answer, the rest of you are pulled off your own path, and items you would have found alone stay buried. The benefit is real too. You hear what you missed, you catch some of your mistakes, and a week later you remember more than if you had spent the whole evening alone.

You do not have to choose between them. Recall alone first. Then talk. Then test yourself alone again.

Two heads together recall less than two heads apart. Two heads that work apart and then compare notes can end up knowing more than either would have managed alone.