Online social remembering

Social interactions, beyond just social remembering, have been completely altered due to the onset of online platforms. In the digital age, people have exponentially more access to the Internet, social networking, and mobile communication, a phenomenon coined the triple revolution (Rainie and Wellman Reference Rainie and Wellman2012). This triple revolution is a global phenomenon, one that has facilitated social interactions at a scale never seen before in human history. People now can connect with others across the globe at any time and exchange information (Marsh and Rajaram Reference Marsh and Rajaram2019). Additionally, the accessibility and availability of mobile devices allow people to be constantly connected to their mobile devices and, in turn, to the Internet and social media accounts (Barnier Reference Barnier2010; Baron Reference Baron2010). Now more than ever, people can connect with others, including strangers, beyond their offline social circles, engage with online social platforms at any moment, and exchange information that normally would not be accessible to them (Chayko Reference Chayko2014). This mass expansion of digital media has also been coined the connective turn, which has shifted memory in numerous ways (Hoskins Reference Hoskins2011).

The digital age has clear influences on how social interactions are facilitated, and consequently, must have implications for online social remembering (eg, Hoskins Reference Hoskins and Hoskins2017; Stone and Wang Reference Stone and Wang2019). Online, or technology-mediated, social remembering can be framed in a variety of ways, making it a rich topic of study across disciplines (Barnier and Hoskins Reference Barnier and Hoskins2018). In cognitive psychology, for example, we may explore if joint remembering on different platforms (eg, text vs. video conferencing) affects memory performance and, if so, what mechanisms are responsible. At the same time, a political scientist might be interested in how online groups translate virtual engagement into real-world mobilisation (eg, Twyman et al Reference Twyman, Keegan and Shaw2017). Likewise, an anthropologist might consider digital landscapes to be areas capable of sustaining broad cultural narratives over relatively long timespans, while a sociologist may focus on how digital artefacts contribute to social cohesion among group members (Agger Reference Agger2008; Pertierra Reference Pertierra2018).

The study of memory in the digital age from a cognitive psychology perspective is relatively new, and it has sparked a profound debate about exactly how the Internet, for better or for worse, has impacted human memory (eg, Carr Reference Carr2008; Schacter Reference Schacter2022). While there is a plethora of studies that have revealed the nature of human memory, and social memory, most of these studies have been conducted in-person outside the context of the Internet and technology (Rajaram and Marsh Reference Rajaram and Marsh2019). Therefore, it is unclear exactly how these studies map onto the nature of everyday memory that often is associated with technology (Storm and Soares Reference Storm, Soares, Wegner and Kahana2021; Wang Reference Wang2019; Yamashiro and Roediger Reference Yamashiro and Roediger2019). Research comparing how memory is studied in the physical laboratory and the expression of memory in the digital world is still in its infancy, motivating many questions about how the Internet influences memory.

Here, it is critical to recognise that there are considerable obstacles that cognitive psychologists must contend with to examine online social remembering. To start, offline memory studies facilitate precise experimental control to help isolate the influence of the experimental manipulation. Virtual studies, however, do not always have that degree of experimental control as participants can be anywhere with access to the Internet throughout the study. Additionally, offline studies typically have an experimenter present to facilitate the session and address questions where online studies do not have an experimenter physically or continuously present to ensure participants are completing tasks correctly. While these are valid concerns with virtual studies that prevent researchers from being able to readily investigate memory processes online, there are also ways to address these issues and conduct experimentally solid studies on virtual platforms (Sauter et al Reference Sauter, Draschkow and Mack2020). For example, experimenters can add instruction checks, add attention checks, and shorten the length of the experimental session. These methods have proven successful in some cases, with some classic studies replicated online (eg, Horton et al Reference Horton, Rand and Zeckhauser2011). The challenges of online experimentation notwithstanding, given that online social networking has become ‘a way of life’, with people having constant, almost unlimited access to online social networks, the time is upon us to examine the properties and consequences of online social remembering. This sentiment has been echoed in recent, related work focusing on the relationship between memory and the self in the age of social media (Wang Reference Wang2022)

Our aim was to ask whether online remembering, both social and individual, encompasses the same properties as the offline counterparts. We briefly discuss the collaborative memory paradigm that we leveraged to create its virtual counterpart, and how its application has contributed to our understanding of social remembering and its downstream impacts on memory. In describing the paradigm, we aim to demonstrate its relevance to research questions in other domains, especially those just described, namely, that the control and flexibility of this paradigm affords insights into the bottom-up processes that shape individual and collective memory.

Collaborative recall and collective memory

Historically, most of human memory research has focused on the individual (Ebbinghaus Reference Ebbinghaus1885). This focus contributed to the development of influential theoretical frameworks that attempt to explain how individuals, working in isolation, remember information (Surprenant and Neath 2009). Bartlett's (Reference Bartlett1932) pioneering work on social memory is an early example of diverging from this path. In many now famous descriptive experiments, Bartlett demonstrated that human memory is subject to social influence. In other fields, such as anthropology and sociology, how societies and groups remember the past has been an active focus of research for many years (Olick Reference Olick1999), whereas in cognitive psychology and cognitive science more broadly, this focus has evolved over time. While influential lines of research have reported on how others may contaminate our memories (eg, Bartlett Reference Bartlett1932; Gabbert et al Reference Gabbert, Memon, Allan and Wright2004; Loftus Reference Loftus1992; Roediger et al Reference Roediger, Meade and Bergman2001), groups and collectives as the unit of measurement have emerged much more recently in cognitive science (Barnier and Sutton Reference Barnier and Sutton2008; Rajaram Reference Rajaram2011; Weldon Reference Weldon and Medin2000).

The relatively recent surge of interest within cognitive psychology in assessing collective cognition can simultaneously draw from and inform research across disciplines (Harris et al Reference Harris, Paterson and Kemp2008; Hirst and Manier Reference Hirst and Manier2008; Wertsch and Roediger Reference Wertsch and Roediger2008). In cognitive psychology, definitions of collective memory have focused on operationalisation in terms of the number of overlapping items recalled independently by all former members of a group. Psychologists have also assessed collective memory with data, showing that people from different countries overclaim the role of their nation for the same historic event such as World War II (Roediger et al Reference Roediger, Abel, Umanath, Shaffer, Fairfield, Takahashi and Wertsch2019). Anthropologists have approached the concept of collective memory at a societal level, examining how groups remember their collective past and how these memories are passed down through generations. For example, Cole (Reference Cole2001) had conducted fieldwork in East Madagascar examining how Betsimisaraka villagers’ social practices preserved memories of their French colonial time period. Similarly, historical studies examine how nations (or cultures) may remember their past (eg, Russia and Georgia's contrasting narratives of the war of August 2008, Wertsch and Karumidze Reference Wertsch and Karumidze2009; Jewish people remembering the origins of their culture, Dudai Reference Dudai, Roediger and Wertsch2022; for a broader commentary, see Wertsch Reference Wertsch, Assmann and Shortt2012) or, in sociology, some researchers are interested in examining how sub-groups within nations come to have contrasting perspectives of their nation's past atrocities (eg, Olick Reference Olick2007; Simko Reference Simko, Roediger and Wertsch2022).

To address the issue of collective memory through a cognitive lens and analyse the bottom-up process that gives rise to shared memory representations, we leverage a collaborative recall framework (for a review, see Rajaram and Pereira-Pasarin Reference Rajaram and Pereira-Pasarin2010; Rajaram Reference Rajaram, Meade, Barnier, Van Bergen, Harris and Sutton2017; for a meta-analysis, see Marion and Thorley Reference Marion and Thorley2016). This approach draws from the collaborative memory paradigm, established in the 1990s by a series of cognitive studies (Basden et al Reference Basden, Basden, Bryner and Thomas1997; Meudell et al Reference Meudell, Hitch and Kirby1992, Reference Meudell, Hitch and Boyle1995; Weldon and Bellinger Reference Weldon and Bellinger1997), where a standard collaborative recall experiment typically entails the following procedure: (1) participants study materials, (2) complete a filler task, (3) recall the study material working individually or in a group, and oftentimes (4) recall the study material again individually. This paradigm offers flexibility for researchers to explore the influence of different study materials, group sizes, and collaboration styles. Moreover, the paradigm can provide insights into how group recall influences subsequent individual memory, affording assessments of post-collaborative memory performance and the emergence of collective memory (Choi et al Reference Choi, Blumen, Congleton and Rajaram2014; Congleton and Rajaram Reference Congleton and Rajaram2014; Rajaram and Maswood Reference Rajaram and Maswood2018). Finally, the collaborative memory paradigm provides a large degree of control over a variety of factors, allowing research to pinpoint the social and cognitive explanations.

Weldon and Bellinger (Reference Weldon and Bellinger1997) were among the first to employ this paradigm in their canonical study. Their key comparisons focused on collaborative versus nominal group recall. Nominal groups – or groups in name only – serve as the baseline and are formed by pooling the non-redundant responses from participants who had worked alone. This provides a fair comparison and is standard practice in this research domain (Basden et al Reference Basden, Basden, Bryner and Thomas1997). One of the most notable findings from this study – of central interest here – was that collaborative groups recalled less than the nominal groups. This effect, referred to as collaborative inhibition, has since been replicated many times in a variety of contexts.

Beyond the collaboration stage itself, collaborative remembering can have a lasting impact on individual memory. Several studies that have implemented post-collaborative, individual recall phases have found that an initial collaborative recall can improve downstream memory (eg, Choi et al Reference Choi, Blumen, Congleton and Rajaram2014; Congleton and Rajaram Reference Congleton and Rajaram2011, Reference Congleton and Rajaram2014). This improvement is larger than what would be expected if the only mechanism at play is hypermnesia – or improvement in individual recall performance across repeated retrieval attempts (Congleton and Rajaram Reference Congleton and Rajaram2012; Payne Reference Payne1987; Roediger and Butler Reference Roediger and Butler2011; Rowland Reference Rowland2014). One explanation for improved recall following collaboration, over and above standard hypermnesia (Payne Reference Payne1987), is re-exposure. That is, group members can strengthen memories for information reintroduced during collaboration. This can act as a second encoding phase, and this improves recall performance in later retrieval recall attempts (Blumen and Rajaram Reference Blumen and Rajaram2008). Some studies show that group members can also impair memory, producing persistent forgetting evident in later individual memory (Barber et al Reference Barber, Harris and Rajaram2015; also see Cuc et al Reference Cuc, Ozuru, Manier and Hirst2006 for an example of such evidence from a different approach).

As we just described, the effects of group remembering reach beyond collaboration, having downstream effects on individual memory, thereby shaping the collective memory of former group members (Choi et al Reference Choi, Blumen, Congleton and Rajaram2014; Congleton and Rajaram Reference Congleton and Rajaram2014). In cognitive psychology, collective memory is usually indexed by the amount of overlapping content recalled by former members of a group (Cuc et al Reference Cuc, Ozuru, Manier and Hirst2006; Stone et al Reference Stone, Barnier, Sutton and Hirst2010). Not only do former collaborators recall similar material, but they also do so in a similar fashion, reporting items in similar output positions (known as collective organisation, Congleton and Rajaram Reference Congleton and Rajaram2014). This additional perspective – examining collective organisation over and above collective memory content – provides a deeper and likely more enduring window into how collaboration synchronises memory for a people. Thus, collaboration facilitates the development of collective memory and homogenises memory structures. In brief, collaboration facilitates the development of collective memory in former group members, such that their memories and the structure of those memories become more aligned with one another.

While the aforementioned effects – collaborative inhibition, post-collaborative recall boosts, and the emergence of collective memory content and structure – are robust in face-to-face collaborative recall settings, we do not know whether collaborative remembering in completely virtual environments would yield similar outcomes. The few studies that include chat-based contexts have reported effects comparable to face-to-face collaborative memory research, including collaborative inhibition, but have found better post-collaborative individual recall in face-to-face groups. This has been attributed to the constraints associated with working in a chat-based environment which, unlike face-to-face collaboration that is laden with non-verbal cues, leads group members to self-filter their responses and/or ignore their partners’ contributions and send duplicates (Clark and Wilkes-Gibbs Reference Clark and Wilkes-Gibbs1986; Ekeocha and Brennan Reference Ekeocha and Brennan2008; Hinds and Payne Reference Hinds and Payne2016, Reference Hinds and Payne2018). Critically, these studies have been situated in lab settings where the use of the electronic chat platform was limited to the individual and collaborative remembering phases. The impact that a fully virtual and remote setting can have on individual and collaborative remembering remains mostly unknown, despite the use of digital platforms that increasingly characterise our educational and social lives.

In light of this background, we explore whether findings from the standard collaborative memory paradigm on the nature of collaborative inhibition, post-collaborative recall, and collective memory for content and structure observed in in-person laboratory settings will generalise to fully online virtual interactions.

The present study

We launch virtual collaborative recall, and report two experiments, to explore individual and collaborative recall in a fully online, chat-based environment. In Experiment 1, our method closely followed the in-person method in Peña et al (Reference Peña, Pepe and Rajaram2021, Experiment 1) in terms of stimuli, design, and task sequence, with few exceptions other than all aspects of the procedure being conducted online. Experiment 2, which included several changes from Experiment 1, was designed to follow up on the surprising effects observed in our first experiment.

Drawing from a rich literature on collaborative memory, we developed several hypotheses regarding classic collaborative inhibition, post-collaborative recall, and collective memory. First, based on the reports of the robust findings from in-laboratory collaborative memory experiments, we hypothesised that Collaborative groups would recall less than equal-sized Nominal groups (eg, Basden et al Reference Basden, Basden, Bryner and Thomas1997). The inclusion of a second recall phase in Experiment 1 allowed us to examine how collaborating with others can shape subsequent memory performance. Consistent with previous work, we hypothesised that former Collaborators would recall more than their Nominal counterparts (Blumen and Rajaram Reference Blumen and Rajaram2008; Weldon and Bellinger Reference Weldon and Bellinger1997). Continuing with the assessment of post-collaborative recall, we hypothesised that collective memory would be greater among former collaborators compared to those who did not collaborate before. That is, we expected to replicate the findings from in-laboratory settings that collaborating would lead to individuals recalling more of the same content (eg, Choi et al Reference Choi, Blumen, Congleton and Rajaram2014; Maswood et al Reference Maswood, Rasmussen and Rajaram2019). Lastly, we assessed whether previous collaborators would have more aligned memory structures as observed in previous studies; here, former group members would not only recall the same information, but they would also organise this information in similar ways when recalling it individually (Congleton and Rajaram Reference Congleton and Rajaram2014).

Experiment 1

Results

In this experiment, we assessed individual and collaborative remembering in a totally online, chat-based context. In doing so, we largely co-opted hypotheses from in-person experiments and expected to replicate laboratory findings. Specifically, we hypothesised that (1) Collaborative groups would recall less than equal-sized Nominal groups, (2) collaboration would boost post-collaborative individual recall, relative to those that never collaborate, and (3) collaboration would contribute to the emergence of collective memory content and structure.

For all analyses in this and the next experiment, we chose a cut-off of ±2 SDs for the exclusion of outliers. Our choice was motivated by the online environment; in the laboratory, when participants can be monitored, relatively fringe performance can be classified as reasonable (ie, by confirming no cheating, no intermittent checking of social media). In our experiments, this was not possible. For each analysis, such exclusions were low and did not change the general pattern of results. For all analyses, alpha was set at 0.05 (two-tailed) and Cohen's d (Cohen Reference Cohen1992) was used to assess effect size.

Recall 1 – group recall

To examine if collaborative inhibition persists in a totally online, chat-based context, we compared the proportion of correct items recalled between Collaborative and Nominal groups. Nominal groups were computed by aggregating the non-redundant recall of three Control participants. An independent sample t-test revealed no significant difference between the conditions, t(28) = 0.17, p = .8661, Difference 95% CI [−5.72 to 6.76%], d = 0.06, d 95% CI [−0.69 to 0.81]. Collaborative group recall (M = 48.74%, SD = 7.96%, n = 15) and Nominal group recall (M = 48.22%, SD = 8.72%, n = 15) were essentially equivalent – see Figure 1. This result – a lack of collaborative inhibition – was surprising and diverged from a large body of work in the collaborative memory literature (Marion and Thorley Reference Marion and Thorley2016). As we note with a reference line in Figure 1, the Collaborative group performance was on par with an in-person study using identical stimuli and the procedure (Peña et al Reference Peña, Pepe and Rajaram2021) and other similar in-person research using categorised stimuli (Congleton and Rajaram Reference Congleton and Rajaram2011). In other words, the online platform did not change the levels of collaborative recall compared to the typical levels reported for in-person studies; the impact was seen in the decline of individual recall levels that constituted the Nominal group performance.

Figure 1.

Experiment 1 – Recall 1: group-level recall (per cent of correct words recalled). Note: The y-axis is the per cent correct, out of 90 studied words. Bars are at mean and error bars are SEM. Individual points and the overlaid distributions represent the underlying data (n = 15 triads of participants in each condition). Horizontal reference lines are from an in-person experiment using the same stimuli and procedure where we observed collaborative inhibition (Peña et al Reference Peña, Pepe and Rajaram2021). The dashed line is collaborative performance from that study, and the black solid line is Nominal performance. Note that Collaborative group recall is nearly even with levels observed using the same procedure in the laboratory, whereas Nominal recall is well below laboratory performance.

Recall 2 – individual recall

We also wanted to see how online collaboration affects memory downstream. We found that former collaborators recalled more correct items (M = 32.49%, SD = 10.57%, n = 45) than those that never collaborated (M = 22.41%, SD = 9.85%, n = 47). This difference was statistically significant (independent samples t-test: t(90) = 4.74, p < .001, Difference 95% CI [5.85–14.31%], d = 0.99, d 95% CI [0.55–1.43]). This finding, depicted in Figure 2, is consistent with previous work, demonstrating that collaboration can boost subsequent individual recall performance (eg, Bärthel et al Reference Bärthel, Wessel, Huntjens and Verwoerd2017; Blumen and Rajaram Reference Blumen and Rajaram2008; Nie et al Reference Nie, Ke, Li and Guo2019; Weldon and Bellinger Reference Weldon and Bellinger1997).

Figure 2.

Experiment 1 – Recall 2: individual-level recall (per cent of correct words recalled). Note: The y-axis is the per cent correct, out of 90 studied words. Bars are at mean and error bars are SEM. Individual points and the overlaid distributions represent the underlying data (Collaborative n = 45; Nominal n = 47). Horizontal reference lines are from an in-person experiment using the same stimuli and procedure (Peña et al Reference Peña, Pepe and Rajaram2021). The dashed line is former collaborative individual performance from that study, and the solid line is Nominal individual performance.

Nominal participant hypermnesia

To further assess the low recall performance during the first recall in the Nominal condition, we analysed the within-subject change from Recall 1 to Recall 2. If low performance was driven largely by a failure to follow task instructions rather than by lowered engagement, we would expect little change between the two phases. However, if recall performance improved between phases, a small improvement of hypermnesia typically observed in successive recalls (hypermnesia; Payne Reference Payne1987), it means that participants did not entirely ignore the task but instead that they were not fully engaged. Indeed, performance improved from Recall 1 (M = 18.91%, SD = 8.09%, n = 45) to Recall 2 (M = 21.80%, SD = 9.61%, n = 45), and this improvement was statistically significant (paired-samples t-test: t(44) = −3.96, p < .001, Difference 95% CI [−4.36 to −1.42%], d = −0.31, d 95% CI [−0.47 to −0.15]). These results suggest that Nominal participants do not perform at their normal standard as they would in the laboratory despite this small boost in performance.

Collective memory – content

Including a second recall phase allowed us to examine collective memory. To assess this, we calculated the overlap in recall, ie, items recalled by all former members of a group (Congleton and Rajaram Reference Congleton and Rajaram2014). Former collaborators, ie, participants who were collaborative group members earlier, recalled more of the same material (M = 10.56%, SD = 3.56%, n = 14) than did members of Nominal groups (M = 2.50%, SD = 2.24%, n = 16). This difference was statistically significant, t(28) = 7.52, p < .001, Difference 95% CI [5.86–10.25%], d = 2.75, d 95% CI [1.71–3.80]. This finding is depicted in Figure 3 and replicates the in-person finding that collaboration promotes collective remembering (Congleton and Rajaram Reference Congleton and Rajaram2014; Peña et al Reference Peña, Pepe and Rajaram2021).Footnote ²

Figure 3.

Experiment 1 – recall 2: collective recollection (per cent of correct words collectively recalled). Note: Collective memory was calculated by dividing the total number of items remembered by all previous group members during their second, individual recall by the number of items possible (90). Bars are at the group means, error bars represent SEM, individual points are group-level proportions, and the overlaid shapes describe the distribution of the data in each group.

Collective memory – structure

We used the Shared Organization Metric Analysis (SOMA – see Congleton and Rajaram Reference Congleton and Rajaram2014 for a detailed treatment) to assess the collective similarity in how former group members organised their later individual recall. SOMA, based on Pair Frequency (PF – Sternberg and Tulving Reference Sternberg and Tulving1977), assesses the number of adjacent forward and backward word pairs reported by previous group members in subsequent, individual recall phases. A score of 0 indicates a chance number of pairs across group members. Former collaborative group members exhibited more similarity in the way they organised their recall (M = 2.07, SD = 1.17, n = 15) than Nominal group members that never collaborated (M = 0.46, SD = 0.55, n = 16). Because of restricted variance in the Nominal condition (scores clustered around 0; Levene's test for equal variances: F(1,29) = 6.64, p = .0153, we used a Welch two-sample t-test to compare conditions, which was statistically significant, t(19.64) = 4.84, p < .001, Difference 95% CI [0.91–2.29], d = 1.78, d 95% CI [0.91–2.64]). This relationship is depicted in Figure 4.

Figure 4.

Experiment 1 – recall 2: collective memory structure measured with SOMA. Note: SOMA scores in each condition, as defined by Congleton and Rajaram (Reference Congleton and Rajaram2014). A score of 0 indicates chance level shared organisation, whereas higher scores indicate greater shared organisation. Bars are at the group means, error bars represent SEM, individual points are group-level scores, and the overlaid shapes describe the distribution of the data in each group.

Experiment 2

Our candidate explanation for the lack of collaborative inhibition in Experiment was that Nominal participants were less motivated to complete the task alone under online conditions and, therefore, did not complete it to their full potential. In laboratory contexts, all distractions would be removed, and the experimenter would be physically present to encourage engagement with the task at hand. These factors do not feature in standard remote, online situations and were not present in the current study for those participants who had to perform the task working individually. We tested this explanation by creating conditions where participants would be more motivated to sufficiently complete the recall task while working individually.

This idea of reduced motivation is similar to the notion of social loafing in which people defer responsibility to others and put forth less effort (Latané et al Reference Latané, Williams and Harkins1979). Weldon et al (Reference Weldon, Blair and Huebsch2000, Experiment 2) tested a possible role of social loafing during collaborative recall by creating conditions in which participants would have higher expectations for their own recall performance and, in turn, would be less likely to coast through the task. Specifically, participants were given information about other participant's ‘typical performance’, which was actually 2.5 times the typical performance the authors observed in their Experiment 1, and were told that they should take their best guess to reach that recall threshold. This ‘forced recall’ task (Erdelyi et al Reference Erdelyi, Finks and Feigin-Pfau1989; McKelvie Reference McKelvie2001) aimed to reduce social loafing and evaluation apprehension (Diehl and Stroebe Reference Diehl and Stroebe1987) by providing participants with a high criterion for performance and encouraging guessing. These instructions improved individual recall for participants in their control, Nominal condition while leaving the performance of participants in the collaborative condition unaffected and still elicited collaborative inhibition in group recall. For present purposes, these results also show a way to increase motivation for Nominal participants.

We built off the work done by Weldon et al (Reference Weldon and Medin2000) to test if lack of motivation influenced low Nominal performance in our first experiment. We made the following modifications to ensure that participants were sufficiently motivated to complete the task to their full potential: (1) participants were given ‘forced’ recall instructions adapted from Weldon et al at the start of the recall phase which provides ‘other participants’ typical performance’; (2) participants were told at study that they would eventually complete an unspecified memory task; (3) the experimenter established presence by re-sending the recall instructions at the start of each recall in the chatroom; and (4) participants studied a shorter word list. Therefore, if our initial null findings for group recall were caused by lack of motivation in the Nominal condition, then we expect to observe collaborative inhibition under these modified conditions.

General discussion

In this study, we launched the collaborative recall paradigm into a completely virtual, chat-based setting. Novel to the current study, we found that remote, chat-based collaboration can facilitate the emergence of collective memory, similar to in-person collaborative remembering. Likewise, collaborative memory performance, in general, was similar to in-laboratory performance and former collaborators exhibited the expected post-collaborative recall boost. By contrast, control recall, that entailed individuals recalling alone instead of in groups, was numerically lower in Experiment 1 than in an in-person study using the same stimuli and general procedure (Peña et al Reference Peña, Pepe and Rajaram2021).

Perhaps, the most striking finding we report is the vanishing – and reappearance – of collaborative inhibition across the two experiments. In Experiment 1, collaborating groups performed equally well as in-person collaborative groups recalling the same stimuli (Peña et al Reference Peña, Pepe and Rajaram2021). However, control performance suffered. In fact, relative to Peña et al and Congleton and Rajaram (Reference Congleton and Rajaram2011), individual recall performance was a full standard deviation below in-laboratory performance. We posit that motivation, generally defined, was reduced in this context for those working alone to perform the recall task, and that this reduction could stem from a variety of sources (eg, task length and lack of supervision), and Experiment 2 supports this idea. We designed our second experiment to address a range of factors that could be responsible for such a reduction (ie, intentional encoding, ‘forced’ recall instructions, shorter word list, and experimenter presence). With these procedural adjustments, individual/Nominal group recall returned to levels consistent with laboratory-based research (Choi et al Reference Choi, Blumen, Congleton and Rajaram2014).

Why would there be a differential motivation reduction between conditions? Given that in both experiments, Collaborative recall was approximately even with in-person performance we have observed in other, in-person studies, it is possible that the social component – being told just before recall that one will be working in a group – increased engagement and motivation. Conversely, when working individually in a remote and perhaps isolated context, there may be little incentive to recall more than a handful of items. This would amount to a reverse social loafing effect. Such an effect would also be exacerbated, though not completely explained, by completing longer and/or more intensive tasks virtually. If task length/intensity was solely at play, Collaborative performance would have suffered as well. In Experiment 2, designed specifically for ease and to afford individuals every opportunity to perform to their full potential, we observed collaborative inhibition. Our explanation of a reverse social loafing effect is consistent with evaluations of the ways in which constant access to online media and digital technology can create memory lapses and distortions where, for example, media multitasking can produce attentional lapses and undermine memory for target information (Marsh and Rajaram Reference Marsh and Rajaram2019; Schacter Reference Schacter2022; Wammes et al Reference Wammes, Ralph, Mills, Bosch, Duncan and Smilek2019)

In addition to providing an approach to assess collaborative inhibition online, we examined the emergence of collective memory. Several in-laboratory studies have established the role collaboration plays in reshaping individual memory, such that, following collaboration, former group members recall more of the same information (Choi et al Reference Choi, Blumen, Congleton and Rajaram2014; Congleton and Rajaram Reference Congleton and Rajaram2014). Experiment 1 extended this finding to a totally virtual, online setting. Together, these findings indicate that while some memory phenomena generalize across offline and and online memory studies, notable differences also emerge across these contexts.

Concluding thoughts