Cognitive Interdependence [Deep Dive] - Moreland Series Part 1

In this post, I hope to describe in more detail a few of the transactive memory studies that were conducted at the University of Pittsburgh and Carnegie Mellon University. Richard Moreland was typically on these studies with Linda Argote also involved in several. These researchers were continuing the series of studies that began with the seminal paper with Diane Liang as lead author that was published in 1995. This study was followed in 1996 and 1998 by other experiments. It was not until 2000 that another TMS paper by this group was accepted into an academic journal.

Richard Moreland had been involved with the transactive memory studies using the electrical circuit tasks since the beginning. He, like Daniel Wegner, was a social psychologist and was primarily interested in how this interdependent view of memory influenced what was known about group psychology. He, with frequent coauthor John Levine, had been extremely influential in the area of groups research. Dick, as Moreland often goes by, and John had both proposed a fairly comprehensive theory of group socialization throughout the 80s that had been widely accepted. The seminal aspect of this theory in the chart below.

Before and after a member joins the group, their level of commitment increases to the group up to a point. At different points in an individual's commitment, they are likely to be accepted, to put in more effort, and eventually to leave the group. Their work after these theory was, to a certain extent, focused on how group members could be brought up the commitment curve faster and be more quickly socialized. This interest, I believe, led the researchers to consider group training and transactive memory as an interesting avenue to explore.

After the initial round of studies, these researchers felt like they had a good handle on the phenomena of transactive memory development. Group members spending time together led them to have more accurate perceptions of expertise, leading the group members to more easily coordinate and trust one another. The manipulation to encourage transactive memory, however, includes more information than just expertise to the group members. It could possibly lead the group member to like one another more because they have spend more time together. A few of the experiments controlled for this factor but the researchers thought that there might be other ways to methodologically deal with this concern.

Enter, Moreland and Myaskovsky (2000). Wegner's theory and the prior papers proposed that the transactive memory of the group is composed of information about expertise. In Wegner's experiments, this was due to the romantic couple spending time together and in the earlier Moreland studies it was due to the group members interacting during the training period. In the 2000 paper, however, the researchers isolated the manipulation to just the aspect that the theory mentions, information about expertise. I think this study is perfect in that is smartly builds on prior work, isolating the mechanism, but keeps many other aspects identical which allows us to generalize the findings to past work more easily.

In Moreland and Myaskovsky (2000), all of the members engaging in the radio construction task worked independently or in a group during that first meeting. Then, for half of the independent groups, their work was systematically graded based on area of ability and compared to the other members. A member would then receive a sheet that said the rank of each group member on each of several different categories of skill. Other groups did not receive this information. The researchers found that just providing this limited information about other members, groups performed just as well as if they had been trained together. This suggests that the information about the member's relative skill is helpful for performance, and as helpful as training the group members altogether. The groups that received this performance feedback were not statistically different from the groups that trained together in their level of TMS as measured from the video tapes. Granted, the groups that received performance feedback instead of training together did perform worse and have lower TMS at a mean level, but the values were close.

This particular study attracted the US Army's attention. These researchers applied for and received a grant from the Army to more deeply investigate the effects of performance feedback on groups, especially groups that have employee turnover (like many army groups do). The Army was interested if transactive memory is helpful in small work teams and if providing individualized performance feedback to the group could be a way of quickly building a team's sense of being a group and performance. I will next discuss these studies (never formally published but available in a technical report).

**Personal information about the researchers was attained second-hand and may not be accurate.

Cognitive Interdependence - Part 4

Competing theoretical underpinnings

Discuss the extensions of TMS more into management, prevalence in that area, the eventual acceptance of Sparrow into Science.

Liang, Moreland, and Argote (1995) sought to bring Wegner's work into wider recognition within both the worlds of management and experimental psychology. Though I am sure that there was other interest percolating in transactive memory in the meantime, Liang's study brought significant added momentum to the research area. As mentioned before, the researchers proposed that groups that had been trained together would perform better than those that trained individually. The researchers found that groups that trained together made about 2 errors on average whereas groups that trained individually made more than 5 errors. When the measures from the videotapes were included, the results were clear. Groups that trained together engaged in more of those three processes than other groups. They coordinated better, developed more distinct specializations in the task, and trusted one another's expertise. The researchers proposed that groups that were trained together were able to coordinate in this way because they had the opportunity to develop transactive memory. These three factors are still considered fundamental components (though some say indicators) of transactive memory within a group. The most widely used scale to measure transactive memory systems within groups was developed by Kyle Lewis in 2003 and measures these three components.

What does it all mean though? The going idea within groups research for some time did not really have a good explanation for why groups typically perform better over time. It was clear that individual perform better over time and that group members grow to like one another over time. However, there were still effects of 'group learning' controlling for these other factors. The researchers made a guess that the development of a shared system for coordination and expertise exchange could help explain how group learning was occurring. After looking into the literature, there was some work in the area of shared mental models but this work suggests that over time groups develop a sharedness in how they think things should be done. The researchers had a feeling though that the reason groups do better over time has more to do with how individuals differentiate, specialize into unique roles. And that is essentially what they found.

To confirm that there weren't other effects, the researchers then did a series of studies that looked at the effects of team building exercises and scrambling team members so that they no longer worked with the same people int he second half of the study. They found consistent results that team-building was not as good as training in terms of group performance. This study helped clarify a secondary point earlier that team-building exercises, though good for some things, do not really help groups perform better. If you are most interested in performance, on-the-job training is much more effective than building bonds with your coworkers. The researchers also found that the effects of training together were not just individual. To test this, they randomly assigned people to groups after they were trained together. What they found was that even if an individual was trained as part of a group, that training isn't that helpful if they are working in a new group. This all suggested that there was something important about group training and keeping that group together.

After these studies came out, Andrea Hollingshead began doing some extremely influential work at the University of Illinois going back to the roots of transactive memory research (starting in 1998). She explored transactive memory using romantic couples and has found some really intriguing effects. Even if a romantic couple can't talk to one another, they are able to implicitly coordinate when given a list of words such that one person remembers a set of words and the other person remembers a different set of words. Hollingshead proposed that if the words fit within one of the person's areas of expertise that they would take more effort to remember it and the other person would know not to commit effort to remembering it.

Lewis's scale, published in 2003, has made the measurement of TMS much easier for researchers. An alternative scale (Austin, 2003) is also used sometime though the difficulty in implementing the scale has led it to be less popular. Transactive memory research is now discussed in many research areas and has been accepted into top academic journals. After a controversial article in Science (Sparrow et al., 2011) it even got a mention on the Colbert Report. As with most scientific phenomena, after the initial flourish, there has been more reanalysis and reevaluation of the phenomenon. Lewis and Herndon (2011) proposed more concrete and systematic ways to think of transactive memory, possibly as an attempt to reduce abuses of the concept by researchers less familiar with its intricacies.

I believe I will publish a few more blog posts on this concept but I hope these four posts have provided a deep and (at least marginally) interesting insight into the origin of transactive memory.

**Personal information about the researchers was attained second-hand and may not be accurate.

Cognitive Interdependence - Part 3

Additional papers
The 1995 divergence
Give a taste of some other work but then say that we're saving it for later.

In the Wegner et al. (1985) chapter, the authors mostly proposed the theory about transactive memory but they did also provide some experimental support. This blog post will try and outline some of the most early experimental work and how the work is interrelated.

The first experiment described in the initial book chapter was published separately as Giuliano and Wegner (1983). In this paper, the researches would bring two people into the lab and have them wrap string around one another. Though this sounds pretty silly, even to me, this process was intended to build a sense of cohesiveness between the dyads. The individuals then answered a series of questions styled after Family Feud where they were asked how they thought others would have responded to a general statement: "A place to sleep" "A place to get pizza" etc. Then the individuals discussed their guesses together, either with the person they wrapped together with the string or someone else. The researchers found that when the couples were 'close' they were more likely to create integrative responses. This just means that the choice they make for the "A place to get pizza" is different than either of their individual responses. This decision to change often came about because the members discussed their opinions and realized that each of them had forgotten some important information. This study suggests that together members were able to be more integrative and critical when they were more familiar with their partner

The second study is mentioned in Wegner's (1987) solo authored book chapter. In this case, and for much subsequent transactive memory research, the authors used dating couples (Giuliano  & Wegner, 1985). In this study, the couples were each given a set of cards with some information to memorize for 1 minute. Then they traded cards with their partner who received 30 seconds with the information. In this way, the participants each received some information randomly that would be easier for them to recall because they had more time with it. Also, the partner got to see the cards second so, while they did not have as much time to look at the information, they could create an internal list of information that their partner should be more expert at. Because the cards had general information on them, it is also possible that one member of the couple may be more familiar with information about computers, history, or popular culture than the other. The author of the study found that if someone saw themselves as an expert, they were more able to recall that information. Also, if a person was not an expert but they received more time with the cards, they also tried to recall more of that information. This suggests that the participants took more of an effort to remember things if they thought their partner was not an expert in the subject.

After these two studies, however, there was not much work done on transactive memory specifically until 1995. This would be a good time to mention time-lag in social psychoogical research. As I mentioned before, book chapters typically have limited editorial oversight. Journal articles, on the other hand, have significantly more with most journals assigning 3 or more reviewers per article. Very generally, once a paper is submitted to a journal, the editor looks at the paper and decides whether or not it is relevant to the journal and of sufficient general quality. If not, it is "desk rejected." The author receives some general feedback but nothing substantial. Though the rate at which this happens varies based on the journal, most journals send out desk rejections fairly quickly (within 4 weeks). This numbers can make the average period of review seem quite short for some journals since they include desk rejections as well as acceptances in the same calculation. Again, though their is variance by journal, I have heard that it can take anywhere from 6 months to 2 years between when a paper is submitted and accepted. Then it may take an additional 6 months to a year for the paper to be published, depending on the practices of a journal. These and other reasons can make the timeline of actual work a bit murkier.

In the early 90s, Diane Liang, a PhD student at the Graduate School of Industrial Administration at Carnegie Mellon University in Pittsburgh, PA came across Wegner's research. From discussions from some of those involved, the experiment that would eventually become Liang et al. (1995) was conceived and completed sometime around 1991. Due to several reasons, possibly the long waiting time for journal acceptance, it took several years until the paper was published. The other authors of this paper would each go on to do very interesting work on TMS that I am sure I will discuss later. Richard Moreland was a social psychologist at the University of Pittsburgh who was interested in groups and training within groups. Linda Argote was a professor at GSIA and was beginning to conceive of and publish her influential line of work on organizational learning. As a component of this work, Argote had explored how learning occurs at the group-level as a possible component of larger-scale organizational learning. Though I do not know who proposed this line of work, the general goal of these researchers was to expand transactive memory to groups as opposed to dyads and to consider how the memory advantages of transactive memory might improves a group's ability to learn and perform. As a secondary contribution the authors attempted to show that training with those whom one will work with and on the task is much more effective than training using another task or with different people. This was proposed as a response to the research on team building exercises which had found little actual effect of their use on organizational performance.

This initial study was quite simple but it sparked a series of subsequent studies and jump started the exploration into TM in management literature. In this study, participants were brought into a lab and trained on how to build an AM radio using an electronics circuit kit. I am told the lab was actually a trailer sitting on the edge of CMU's campus though these trailers have since been removed. This kit consisted of a lot of small parts such as transistors, resistors, etc. The first time the group came together, they trained on building the circuit. One week later they returned to perform building the circuit. In one condition, the group was trained all together. In the other condition, each person trained separately. In the group training conditions the members could talk, strategize, and plan on dividing the task which those in the individual training condition did not have the opportunity to do. From video tapes of the group's performing, the researchers rated the groups on how well they were working together on three different factors: coordination, memory differentiation, and credibility. Based on Wegner's theory of transactive memory, the researchers proposed that these three qualities would be more prevalent in groups that had developed a TMS. The researchers were interested in 3 effects. First, did training condition directly influence performance? Second, did training condition influence the extent to which the group members engaged in those three factors: coordination, differentiation, and credibility. Lastly, did those three factors predict performance better than the training manipulation?