Team Situation Assessment and Information Distribution HENRIK ARTMAN1 Division of Operational Studies, National Defence Collage Sweden [email protected] This paper describes the cooperative work of constructing team situation awareness within two teams of a military command and control unit. Specifically we discuss how the distributed cognitive and cooperative work of decision-making of the two teams is structured. The situation enabled two different ways of distributing information within the team: one serial and one parallel. One team chose the parallel information transfer, the other the serial one. We discuss the interaction patterns that emerge in the respective teams and their consequences for situation assessment and situation awareness. We then conclude by discussing the differences in terms of means of sharing information. We also offer some hypotheses for future research. Keywords: Dynamic Decision Making; Command and Control; Situation Awareness; Co-operation 1. Introduction The command and control unit under investigation is a military staff unit at battalion level with the task of ensuring that its company units can control the enemy forces. Its work is to plan and transmit orders to its subordinates and artillery units, and also to construct an overall picture of the situation at the front, in order to be able to direct and coordinate action between troops and companies. Furthermore they should be able to forecast what the enemy will do next and how to best utilise its own forces both now 1 Most of the work has been done while affiliated with the Department of Communication Studies, Linköping University, Sweden. and in the near future. Thus its task is to assess the emerging situation, or in other words produce an appropriate situation awareness of the front, and ensure that its own forces may be able to hinder unwanted developments. The staff must gather information from several units in order to understand and predict the situation, i.e. produce situation awareness of the whole battle front which includes both their own forces and the enemy forces. This paper deals with how technology arrangements enable distribution of information, and analyses how this may affect the situation assessment process. Technology is, according to a socio-cultural theory, tightly coupled to cognitive functioning (Hutchins, 1995; Wertsch, 1991). From this point of view it is the combined effort of humans and their machines, or more generally artefacts, that counts. Humans have the privilege of interpreting the meaning of information (Salomon, 1993) but their interpretation will always be affected by how different technologies represent the information. The human-machine cooperation perspective entails many aspects ranging from socio-historical to physical ones. Here we will discuss the physical distribution of information, which is enabled by two different forms of technology architectures, that is, how the technology is arranged to allow information flow between agents. Thus the team is considered the primary cognitive information-processing unit, rather than creating a division between actors, practices and technology. Both experimental and field study research suggests that the way in which information is distributed within a team affects its performance (see Hutchins, 1995; Stammers & Hallam, 1985). Brehmer & Svenmarck (1995) have experimentally investigated how geographically distributed decision-makers, together considered as a team, manage to control a dynamic system. They compared an information distribution using a hierarchical architecture and one using a so-called fully connected architecture where everybody can communicate with everybody, and found that the hierarchical condition managed to control the system better, but only marginally. However, in their experiment, the central, commanding person in the hierarchical condition is alone responsible for compiling adequate information and assessing the future of the system. This may cause a too high workload because of the limited cognitive capacity of individuals. Many other coordination centres (see Heath & Luff, 1992; Hutchins, 1990; Suchman, 1996) have divided the overall task of situation assessment into several more simple tasks which are distributed among several individuals. In this military case one person is responsible for handling the companies and their information processes and another person is assigned to fight enemies with artillery fire. The former mainly coordinates actions that have to be performed in conjunction with several units and keeps track of where the forces are, while the latter mainly tries to control the enemy by using the artillery forces which the battalion and the brigade have at hand. This division of simple tasks calls for distribution and 14 coordination of information at the local level of the team in order to construe appropriate situation awareness. 2. Team Situation Assessment In the relatively new research field of team decision-making several researchers (Klein & Thordsen, 1989; Orasanu & Salas, 1993, Duffy 1993) argue that teams differ from general groups in that the members explicitly have different roles and tasks, and thus attend to different items of information in the decision process, while members of groups in group decision-making often are fully involved in the decision process (see also Brehmer, in prep. for a critical discussion). Furthermore, teams are more often thought of as being in a more elaborated context with special artefacts and tools than the generic group in experiments. In group decision research the emphasis is on how a group reaches consensus (Brehmer, in prep). This is not as important in team decisionmaking as all the members of the team have well-defined roles and tasks and the goal is more one of merging the information together with a common goal than of making a decision that all accept, it is argued. In the team, one member may concentrate on gathering information, another on analysing the information but they may not have to communicate about the actual decision since this is a third person's responsibility. This division of labour has consequences for the team’s cognitive functioning and thus for both situation assessment, situation awareness and decision-making. During the last ten years the concept of situation awareness (SA) has become more and more important for the control of dynamic systems (see Endsley, 1995; Vidulich, Dominguez, Vogel& McMillan, 1994), but only a few researchers deal with Team SA (Salas, Prince, Baker, Shrestha, 1995; Wellens, 1993; Samurçay & Rogalsky, 1993). Even though adequate SA is not itself sufficient for good decision-making (Flach, 1995) a well-informed and well-anchored decision (planning, commands etc) has better chances of being accepted within the team and down the hierarchy. Thus SA becomes an important concept. Wellens (1993:272) defined group SA as “the sharing of a common perspective between two or more individuals regarding current environmental events, their meaning and projected future“, and Salas, Prince, Baker, Shrestha (1995:131) defined team SA as ”at least in part the shared understanding of a situation among team members at one point in time”. Although we do not fully disagree with these definitions we would like to stipulate a definition that puts more emphasis on the interpretative and distributed nature of team SA; “ Two or more agents active construction of a situation model which is partly shared and partly distributed and, from which they can anticipate important future states in the near future.“2 This definition puts more emphasis on the process than the state. This because the state of situation awareness is not simply the sum of individual SA or a completely group level idea of a situation, it is an actively communicated and coordinated accomplishment between several members. This accomplishment emerges in a context where artefacts and information technology partly structure the possibility of sharing and distributing information. Team SA is related to the concept of a shared mental model (Cannon-Bowers, Salas, Converse, 1993). Shared mental models include the team members’ models of the coordinating routines and knowledge within the team, while, SA is the conception of the situation ‘out there’, here the battle field. Cognitive functioning of shared mental models is vital for how the team collectively assesses the situation, as the means of sharing information is what the team coordinate their efforts by. Teams that work closely together learn to know each other’s needs, knowledge and tasks and can adjust their information sharing and interpretations to those needs, a situation we have called cognitive empathy (Artman & Wærn, 1998). Cognitive empathy means that someone with a shared understanding of how to coordinate does so in an appropriate way without having a shared history and complete understanding of the other persons situation awareness. Thus, being able to take the appropriate actions without being told to do so, or having explicit cues for doing it. So it is more than a predictive or formal model for the coordination of work. 3. Technology and Information Distribution Most human-computer interaction research focuses on the internal processing of information by artefacts or humans. Here we discuss information distribution and technology assisted communication as the arrangement of equipment within the team. In Hutchins’ words The arrangement of equipment in a workplace might seem to be a topic for traditional, noncognitive ergonomics. However, it has an interpretation in terms of the construction of systems of socially distributed cognition. The interaction of the properties of the senses with the physical layout of the task environment defines possibilities for the distribution of access to information.(Hutchins, 1995:197) This article highlights the possible consequences of distributing information in different ways in the social system, and how interpretations are sequentially constrained by dif2 16 This definition is informed by Hutchins (1990) and Weick & Roberts (1993). It is discussed in more depth in Artman & Garbis (1998). ferent transformations (Hutchins, 1995). First we want to describe how the two teams distribute information by means of the technology. Then we want to discuss the possible consequences of this difference in distribution. How are the teams structured in terms of information flow and means of coordination? What difference do different information flows imply for the team situation assessment and awareness? First and foremost this case study is intended to be hypothesis generating. We wanted to start from a work situation where actual practitioners used their ordinary methods in order to be able to work out hypotheses, which we can test in an experimental situation. The research questions concern team decision-making and particularly how practitioners organize their work, given the kind of artefacts they have at hand. Here we will look at two organizations that emerge in two military command teams. The organizations emerged spontaneously, without our intervention, during training in a simulator, and originate from a very small change in the technology. This minor alteration may have major consequences. Let us first make our acquaintance with the team roles and their responsibilities. 4. The control room The battalion's overall goal is to delay and prevent the enemy from invading the territory. In order to do this the focused team has to coordinate its forces and the brigade artillery. The team we will focus is in the middle of a command chain, where they receive general goals from a supervisor major situated at another place, and deal with requests from subordinate units. The teams work is thus to mend these demands and prioritise actions to reach the overall goal of defending the assigned territory. The team members have quite distinct work assignments, though the outcome is dependent on the team's collective achievement. A full battalion staff consists of two teams working in shifts. The room where the staff is housed is a small tracked vehicle identical to the kind used in an actual combat situation. It is about 6 metres by 2.5 metres in area, with a big table in the centre where the staff keep all their equipment. The staff consists of four people, two signalmen and two officers. The signalmen sit at each end, and the officers in the centre, see Fig. 1. FIGURE 1. The control room setting. The signalman (S1) sits at the rear, the coordinator (C) in the middle and the artillery leader (A) up front. The other signalman (S2) sits just under the camera. 4.1. Staff and responsibilities The signalmen (S1 and S2) each have a small computer for which they are responsible. They write and receive messages from hierarchically subordinated units. S2 sometimes also communicate with supervisor units by writing acknowledgement of actions that are of broader interest (such as brigade artillery fires). They have each a small printer on which they print out the messages and distribute them to the officers within the team. Each officer has specific tasks connected to his rank and formal competence3. We call them here Coordinator (C) and Artillery leader (A) respectively. The Coordinator is responsible for the coordination and communication with the distributed forces, as well as for the accurate updating of the map. His main task is to gather and analyse, information and then update the map and distribute relevant information (situation reports, orders etc.) to appropriate units. The Artillery leader's task is to decide how the artillery resources are to be used and which unit should use them. This means that he has to analyse, prioritise, issue orders and inform the units about when and how much artillery they can use, as well as to order the artillery units to fire and where and when. At the same time he must inform the supervisory battalion major, situated in an another tracked vehicle, about the actions. To make a decision he uses the information the Coordinator has gathered and organized. Together the officers are responsible for the planning of and predicting how the dynamic situation could change. 3 18 We use here the pronoun ”he” as all the team members are males. 4.2. Artefacts In front of C and A lies a map. They also use a voice-radio system to keep in contact with units outside of the staff. The map is the main mediating tool combined with verbal communication, for C and A. The map looks like an ordinary map with a coordinate grid and general signs (there might be special details for military purposes) of the area for which the battalion is responsible. As has been noted above, C is responsible for updating the map with symbols relating to the movement of his own forces and the enemy forces; for this he uses different standard colours and symbols. This map has two general purposes. First, it is used as an overview of the battalion and the enemy forces about which they have information. It is from that information they decide about movements of the forces and the use of artillery. Second, it is used as a tool for discussing and hypothesising about what the enemy is up to and what they themselves can do with their forces to hinder them. The map is a physical representation of the team’s situation awareness, but it does not visualise the dynamics or the processes of the battlefield or show items of information that C has not recorded. The team receives all new information by a voice-radio or by e-mail, and occasionally directly via other members of the battalion who visits the vehicle. The voice-radio console includes two radiophones to contact distributed units. One phone has loudspeakers so everyone in the room can hear what the persons at the other end are saying. This is of course very important as it provides the members of the team with a common ground on which to discuss and decide about further actions. This loud speaker system is mainly used by C, to seek and receive information from the distributed units. The other phone is a "private phone", which means that the other team members do not hear what the other person is saying. This phone is connected to a tracked vehicle where the supervisory battalion major is located. To keep track of what is said on the phone they also have a logbook where all spoken messages by radio are to be entered. Just prior to the data collection Swedish army command and control units had been equipped with a new, complementary, tool for communicating with distributed units. It is called DAta Reporting Terminal (DART). In simple terms it can be described as a notebook computer designed for handling electronic mail. Each command and control unit has two such DARTs, one for each signaller. Each DART has a printer and also has a memory in which messages can be stored. The DART can be programmed with certain formats to ease and standardise message writing. If a programmed formula is not wanted messages can be typed in clear, which can be compared to an ordinary e-mail. The printed messages are then given to the officers who take care of the written messages and then put them on pins for incoming or outgoing messages, thus providing a chronological logbook of the messages they have received and sent. The messages are also time-logged. This logbook can be regarded as a collective memory of what has been done. This is useful if C has not updated the map properly or if they have to check specific messages again. 4.3. Procedures that must be accomplished Quite schematically we will here try to describe what the team has to do. The command and control unit receives a message from the subordinates, either via the radio or the e-mail system. If the message is received by e-mail it is confirmed by one of the signalmen, who then prints it out and gives it to C. If the message is received via radio the Coordinator takes care of it directly. In both cases C evaluates the message and transcribes the information onto the map. When the information is on the map it constitutes, together with other information, a second time-scale with respect to when the information was gathered. If the message is of importance for the artillery (for example if a company or troop calls for artillery fire) C gives A the message, else he sticks it on the incoming pin. When A has received the message, he reads it and evaluates what, if anything, the artillery can do. Often there are several messages calling for artillery fire and then A has to prioritise. When he has decided on artillery fire he must send a message to the artillery (either subordinate, supervisor or both). He tells one of the signalmen to prepare for such a message and then dictates it to the signaller, who also transmits it to the originator of the request as well as to the supervisor major. Every now and then A and C have to confer about what the enemy is doing and how they can plan in order to be one step ahead and delay or stop him. As we will see, how these procedures are carried out in the teams is correlated with the technology architecture, i.e. how the DARTs are connected. Before we can discuss the direct work we must, therefore, have some idea of how the technology can be configurated. 5. Technology architectures The two DARTs can be connected to other units in at least two different configurations. One of the architectures lets one of the DARTs be connected to subordinates (hierarchically below) and the other to supervisor units, such as the unit where the battalion major is situated or to battalion commanded fire units (hierarchically above) (see Fig. 2). The other architecture lets both DARTs be connected to the subordinates and only one to the supervisors (see Fig. 3). It is possible at any time to shift between the two architectures. In both cases it is possible to send messages between the machines via the so-called local link. 20 Subordinate S1-Dart Local link Supervisor S2-Dart Fig. 2. Shows how the technology architecture when one computer is connected to each of the lower units Subordinate S1-Dart Local link Supervisor S2-Dart Fig. 3. Shows the technology architecture when both computers are connected to the lower units The rationale for having these two structures is that if the S1-DART breaks down all messages can still be received by the team. This makes the means of communication more robust. In both structures the signalman (S2) operating S2-DART should transmit information both to the subordinate and the supervisor. In the architecture depicted in figure 2 he must then transmit via S1-DART, while the architecture in figure 3 allow direct communication. The direct communication mode although also make that all information from subordinate units reach directly to the S2-DART, which then S2 has to deal with. Using either architecture poses different opportunities and communication structures within the team, these communication structures in turn has consequences for how situation awareness is distributed within the team. 6. Method The present study has used video-audio recordings and computer logging as the main data collection resource. We video-recorded a full battalion staff training session. The training was divided into two parts, the first lasting from 0830 to 1930 hours, the second from 0800 to 1100 hours, in total fourteen hours. One of the teams used nine hours of the training session. The two teams took over each other’s combat situations every 2-3 hours, so they have worked on the same overall problem but within different temporal phases. This means that there could be some differences in workload, but as always in dynamic decision-making situations are not exactly comparable as a consequence of the fact that the dynamic system changes both autonomously as well as by the control system’s actions (Brehmer, in prep.). We could not be present in the control room because of lack of space and the possible disturbance factor. Instead we observed the sessions from a monitor connected to the video and could ask other officers what the teams were doing. We have transcribed 4 hours of sampled chunks of the video recording. Many types of messages were sent to the teams, and one type was of relevance for all team members, i.e. they had to process it in one way or the other was an artillery request that had to be coordinated with the supervisor battalion major. Other types of messages only address one or two persons. As for the criteria by which chunks were analysed, we sampled eight messages out of 71 of this specific type of message. We transcribed communications in half an hour prior to the chosen message in order to see the context in which it was processed. This was done in order to see if there were any recurrent negotiational processes between the team members. By this procedure we covered 28 (39.4%) of the 71 messages. One team processed 20 of this type of message and the other sent eight messages4. This reflects the proportion of time the teams spent in the session. It was found that the teams used the technology differently, which was not known while first analysing the cooperation. The transcriptions originally included both what is said, who is talking to whom (or to ”the room”), actions such as gestures and other comments about certain matters, such as how to use the different representations etc. In the translation from Swedish to English we have simplified the excerpts by excluding pauses, addressees, non-task relevant actions. The analysis started by looking at how the teams differed in respect to cooperative work. It was evident that the two teams chose to work in different ways. They did so deliberately, we did not tell them to do so. We traced the information processing and analysed how the mediating tools and interactions made the work coordinated. The different work patterns that emerged for this analysis elicited several questions. We therefore analysed interactions in the teams’ work to show how the technology might enforce different situation awareness between the team members within the teams, but still withholding the prescribed taskallocation. Thus the excerpts in tables (2-4), in section eight in the article, are representative of 28 interactions with the focus on this type of message. 7. Interaction patterns As we have seen the technology can be configured in two different architectures, one where subordinates are connected only to one DART (Fig. 2), and the other where both DARTs are connected in parallel to all subordinates (Fig.3). This difference gives rise to different demands on the team and different forms of information distribution. We call these two emerging interaction structures serial and parallel respectively, as the informa4 22 When starting to analyse this data we sought negotitional aspects of team decision-making. We did not know that the team’s worked differently from the beginning, this became evident as we transcribed the randomly chosen chunks. Thus, the method owes for other research purposes and therefore the chunks are unevenly distributed between the groups. tion is processed in such ways. Under normal circumstances (e.g. when both DARTs are in use), the teams are taught, to work by using the serial technology architecture. Despite this, the teams chose to organize their work differently, one team worked with the serial architecture while the second worked with the parallel one5. The interaction patterns that emerge from the use of the technology adhere mostly to a mild form of Schmidt’s (1991) distinction between collective and distributed cooperation. Collective cooperation is overt and conscious, while distributed cooperation is divided and semi-automous. In this case each individual has his own tasks, thus the cooperation is distributed to some degree, but using the serial technology architecture enables or affords more collective cooperation, than the parallel technology architecture. We follow the information flow within the team and highlight the stages when the information is transformed into different representation states. Bear in mind that we are concerned here only with information received via the e-mail system, and not when information is received via the phone. 7.1. Serial work organization; Team 1 This team is less stressed than the other team. They work calmly and methodologically through all incoming information. This group seems to be quite socially oriented, i.e., they talk quite a lot with one other and try to resolve most problems collaboratively. They use the technology as depicted in figure 2, where each DART is connected to the subordinate and supervisor respectively. If we follow an average message the serial organization works as follows: 5 There is, for us, no possible way to tell why they chose different procedures. It can be that one team had better understanding of the technology and what could be done with it. In any case we have found in laboratory experiments that people are creative in this way, and do not always act as they are supposed to. Subordinate Supervisor 6 to supervisor to subordinate 1 6 S1-Dart S2-Dart S2 Sends 2 Gives Gives S1 3 C 4 Dictates A 5 Says 7 FIGURE 4. Shows the serial interaction pattern when using the normal technology structure. The numbers denote the stages of the information processes. S1 monitors his DART to see when an incoming message is received. When the DART tells him that a new message is received he prints this message on a small printer (Fig 4: stage 2). When the message is transformed into a paper strip it is more easily handled within the team, and can constitute a buffer of information as S1 lays it in front of C, who then can deal with it as soon as he has time (Fig. 4: stage 3). S1 is repeatedly also told to tell C when the message seems to be an urgent one, and he also sends an acknowledgement to the originator of the message. C then picks up the strip of paper and evaluates it. Often this is done aloud, as if he is talking to himself. Sometimes A is then able to hear what the message is about and then promptly helps him out and prepares his own task with respect to this message. C transforms the message to symbols and colours on the map. When the information is on the map it is potentially accessible to all members and constitutes a second-order time-scale, in respect of where the information originated. From this map the team members can read out possible ways of reacting to the situation. Coordination between C and A, who have different aims with the same information, is mediated both verbally and by paper strip and by map. As A is dependent upon information that C distributes he is constantly monitoring C’s work and information, which helps at all times to update his mental situation awareness in concert with C’s. This monitoring and eavesdropping is a very important coordination mechanism (see Heath & Luff, 1992). Much of the Team SA is represented, and thus distributed by means of the map. After C has noted the information on the paper strip he puts it down or promptly gives it to A, who evaluates the message in accordance with what he can read from the positions and symbols on the map (fig 4:stage 4). This evaluation is often done silently. 24 If there are several messages he also has to prioritise the order in which they should be handled. C and A could start to discuss the meaning of the messages and thus interrupt the fluent information flow. We will look at such discussions later. Whatever the content of the message, he has to send an acknowledgement to the originator, so he dictates a message for S2, who converts it into electronic form (Fig 4: Stage 5). When this is done S2 sends the message via the local link to S1 and to the supervisors (Fig 4:stage 6). In order for S1 to know what to do with it (and not just print it out automatically) A tells him to re-send (or let it through) to the subordinate originator, as S2 does not have access to the subordinates (Fig4:stage 7). Then the decision sequence for this particular message is covered, and several sequences have started. C is central as regards information gathering, but A is central for getting the information through. In this serial processing procedure, A becomes quite central as he relies on information from C and S1 on his left, and must also tell S1 to re-send the message going back to the originator. "A" thus has to monitor outgoing messages to ensure they are forwarded. The information is redundant at several levels but cannot be dealt with in parallel as the communication is sequentially organized. It would not happen, except by mistake, that A would get information before C and that he then would make a decision on information (e.g. the map) that is not properly updated. C will always update the map before A gets this particular information. In that way this organization is dependent upon C’s interpretations. But the information processing also takes time; and time is essential in controlling dynamic systems. 7.2. Parallel work organization ; Team 2 If the other team was working calmly, this team is working hastily. They seem to try to process the information as quickly as they can and use the technology as efficiently as possible. Instead of writing a new message to send to the supervisor and subordinates, this team re-uses the information in the electronic message they received, and so save some minutes per decision made. In the beginning they switched between the two technology architectures described above. Their information processing procedure started as it did in the other team, but when A had made a decision he told S1 to send the message received via the local link to S2 (Fig. 5: stages 5-6). A and S2 could then elaborate the message on their side of the room (Fig.5: stage 7). When they are ready A switches over to the parallel technology architecture (Fig5:stage 8-9), which enables them to send the message directly to the originator, without sending it via S1. Thus the team makes use of the opportunity to modify at any stage the electronically held information. The procedure is depicted in Fig. 5. Subordinate to s u b or d i n a t e 1 S1-Dart Supervisor to s u p e r vi s or S2-Dart 2 S2 8 6 Op e n 7 Dictates 3 4 S1 C 9 A Sends Gi ve s Gi ve s Sa ys 5 FIGURE 5. Shows the interaction pattern when alternating between the two technology architectures. The numbers denote the stages of information processing. In order for this cycle to continue, the team has to switch between the two technology architectures every time a message is sent. In fact, on almost every occasion they forgot to change the technology architecture to serial, which made incoming messages reach S2 directly. It is also interesting to note that A seldom or never tells the others that he switches between architectures. This initiated a new form of interaction structure, which the team adopted after a while. We therefore abandon the above structure and move to the more extreme structure that the team adopted. When the technology architecture which connects both computers to the subordinates is used transmission via the local link is unnecessary, as all information received from a lower hierarchy is directly duplicated to both DART-computers. Subordinate 1 to 1 Supervisor t o s u p e r vi s or 2 s u b or d i n a t e S1-Dart S2-Dart S1 3 Gi ve s 26 S2 3 Gi ve s 2 C A 4 Di c t a t e s FIGURE 6. Shows the parallel interaction pattern when using the parallel technology architecture. The numbers denote the stages of information processing. The team chose to organize their work as depicted in Fig. 6. As there are fewer information processing stages for each message, and as they can re-use the information contained in the received message, and thus do not have to manually coordinate as much, this team processes information more quickly than the other team. In this interaction structure S1 and S2 mirror each other to a great extent; both print out messages and give them to C and A respectively (Fig. 6 stages 3). C transforms the information onto the map, but instead of forwarding the message to A, as in the serial team, he sticks it directly on a pin, as a kind of logbook. A who receives all incoming messages from S2, then consults the map and sometimes C, evaluates and prioritises on his own between the messages concerning his task. Then A tells S2 what to change in the incoming message and lets him send it to the originator (Fig.6:stage4). Thus, using this technology architecture A does not have to receive any paper strip from C, nor do they have to talk in order to attend to specific information (or paper strips). The interaction between A and C’s information requirements are mainly managed via the map they have in front of them, thus the interaction between A and C is reduced to almost nil for their routine work, and A’s need to monitor C’s work is nil as he receives all information from S2. A must concentrate on filtering the information himself. The process of gathering and organizing the information is distributed to two subteams. The map becomes an even more important physical representation of SA, as A is using the information on his own without consulting C most of the time. This procedure may, as we will see, mean that C’s and A’s mental models can come out of synchronisation as the updating of the map (C’s work) and the use of the information contained in the map (A’s work) is not sequentially constrained. The only time they have to interact is when they want to discuss more strategic problems or hypotheses. As we will see next, such discontinuity in cooperation may be a problem. 8. Assessing the Situation These two forms of work organization, we propose, have implications concerning how the teams coordinate and construct a situation awareness model of the front6. The task is 6 Please observe that we here are considering the negotiational phase of situation assessment. The assessment is mainly done by the commanders C and A. The full procedures to actually send out the commands is accomplished as depicted above i.e. by sending a message to con- naturally time constrained and there is a trade-off between making decisions upon information which is as complete or adequate as possible and making a quick decision. We have seen how the two teams cope with this in different ways. The team members of team 1 continuously talk more with one another, while team 2 work steadily with individually assigned tasks and choose the technology that supports such an organization. We suggest that these differences may show differences in the participants’ shared representations of the battle situation (see also Grusenmeyer, 1995 for a similar discussion). The information distribution, enabled by either serial or parallel connected DARTs may explain the dramatic differences in hypotheses about future development generated by the two teams (see table 1). TABLE 1. Shows the absolute number of analyzed messages and generated hypothesis in the two teams. Number of analysed messages Number of hypothesis Team 1 (serial) 8 8 Team 2 (parallel) 20 1 Consider the following illustrative examples of interaction in the assessments of the situation in tables 2-4. Both examples occur at the beginning of the training and both are interactions between A and C arising after a period of low workload and when a new message asking for artillery fire arrives. TABLE 2. TEAM 1 (SERIAL). (simplified transcription)7 Who Message C A C S1 C 7 28 Where did we have the support platoon? Was it about here? Yes Because, we have now here a smaller troop with about 80% losses. RN [a company] assess this enemy to be destroyed, but there is a lone army cabin , firearms and they open fire from here, and they could have had a firetroop around here, huh But the best is that there is ammunition about here Yeah, with destroyed food and clothes, or what ever... Action Points on the map Nods Points on the map C and S1 nods Points on the map firm or disconfirm a request, or by sending a command to a specific unit. S1 and S2 are involved in the full procedure (receiving to sending messages) . The excerpts here are simplified which means that we have excluded pauses, intonation and addressee. Clarification by us is within []. A C S1 C Yes, The question is, could we put, ehh, It’s too early to Points on the map put this here... The question is what happens when SN [a company] gives a report, because he must have e some knowledge about that Yeah, that means for X that It was from RN, was it? Yes, it was from RN, but that means for X that it is not safe for him to drive there Points on the map In excerpt 1 C, A and S1 are discussing equally elaborated ideas about the future state of the situation, which is a rare occasion in the other team. They form a hypothesis about what might happen and what the enemy will do next, by using the map as a conversation piece. The map contains more information than one person can actively hold in their short term memory, and they point emphatically at the map to stress possible developments of events. We can see how they actively try to construct the development by forming a hypothesis and anchoring these on the map. C, who is the one who has the responsibility to update the map, even asks C about the state of things. They try collectively to assess the situation. Below, in excerpt 2, we see the only situation in Team 2, which includes a hypothesis. Instead of being collective, the dialogue between A and C is concerned with getting A enough information in order to be able to do the job. TABLE 3. TEAM 2 (Parallel). (simplified transcription) Who Message C [C gives a message to A] A No, what the hell should I say? WHAT THE HELL SHALL I SAY? It must be these (pointing at the map) who have moved or C Hmm, yeah, it must be yes, they are dividing now yeah and trying it way up here A What does that mean? It means like, like it looks like C I think the enemy’s simply trying a way where he does not meet resistance, huh A And as here it is quite a big front here C C Action Points map on the Points map Points map Yeah he has sent [inaudible] He goes here with different Points battalions and companies and tries to move forward and map seek resistance. He should meet one hell of a lot of resistance because here is actually a battalion (hmm) SO he will probably not get through here (hmm) There is a battalion [the same] and here there was a brigade or battalion or? [no answer] This 49 I think I can delete like (another person enters and the conversation ends) on the on the on the A asks C questions but does not, as A in the other team, give his interpretation about the current situation. These questions could at first glance be seen as critical but informed inquires, but is articulated in a more desperate voice. Also note how uncertain A is about how to interpret the situation from the map. Instead the thread is developed by C in order to inform A about the status of the map. This is the only observed hypothesis about the future state of the system that this team makes together in contrast to 8 in team 1. They have supposedly constructed different models of the situation, and must resolve this difference. It is C who communicates the most developed model. A mainly questions C to get an situation report. In Team 1, which uses the serial work organization, the map is always updated before the decision about artillery is made, whereas in the Team 2 parallel organization it is quite possible that A could receive and decide about a certain message before the map is updated, and thus make a decision on older information. Thus, the information flow gets out of synchronisation. Still they work as if C has got the latest information. In the above transcription from team 2 we can see that A does not know exactly what to do. A does not have, mentally, the information from which he could start to assess the situa- 30 tion properly. A contributes by giving questions to C and by that are able to adopt C’s situational model. A is trying to gain the uttermost from C’s situation awareness but does not contribute to the collective situational model himself. This sharp task distribution and difference in situational assessment in team 2 is even clearer in the next episode where A has not noticed a message of concern to him. C promptly makes him attend to this message, but A does not understand in what context this message comes, i.e. what is meant by it. TABLE 4. Team 2. (simplified transcription) Who Message C A C A C A C A C A C Have you seen this? He wants to shoot at 225,59 Does he want to shoot now? Uh Huh, What happens? But wait, you have to tell me what happens! Quick orientation! Yes, they want to cut this way here by the bridge ”I will send values [coordinates]” What does he mean? Where does he want to shoot? Why? He will send the coordinates Uh Huh, he will send the coordinates I hope I suppose he has not said that... he should He has not said something about firing there now? No, he did not You’ll take care of that won’t you? Yes Action Points on the map A reads the message aloud Points on the map We can see that A does not understand the message and what it signifies, and C who does understand, has a problem in updating him as the message must be put in a long chain of messages in order to make sense, and A has not followed or heard of this because he has been concentrating solely on his own tasks and only selectively picked up information from his DART-computer. By not attending to C’s work A has lost the overall awareness of the situation and therefore must ask C about a ”quick orientation”. Again we can see that A does not contribute with own interpretations, but only questions C about his situational model. Taking this case as illustrating the different forms of cooperation, we interpret it as the teams using the parallel technology architecture after a prolonged period lose contact with each other and therefore also shape different situation awareness of the front. Furthermore, as they get out of synchronisation as to their respective situation awareness, the shared mental model of teamwork is no longer ap- propriate and they have to restore the collective situation assessment. The map, which is constantly updated and materializes the situation at the front, is not of much help to A as the information here also must be understood in its historical context. 9. Conclusions The teams share operative information by at least three means; artefacts, attentive monitoring and negotiation. The different uses of the technology, we propose, enables or affords different degrees of the coordination means. We suggest that the arrangement and use of the technology makes the means of information sharing differently weighted in the two teams. Each arrangement pose different valuable trade-offs, i.e. between speed of decision process and well-synchronised anchoring of the decisions within the team. Using the technology to support serial information processing as team 1 did, C will always receive any information before A and A will always be dependent upon the fact that C is passing information to him. The tasks and information flow are sequentially constrained. This dependence allows A to continuously monitor C at any ”free” moments and follow the propagation of the information, and in turn prepare his individual tasks. When A receive information he has already seen the changes in the context and the decision environment has been documented. In team 2 the cooperation is more discontinuous. C is working by updating the map, and A is fully occupied in filtering information from his parallel-connected DART in order to find the information that is relevant to him. There is simply very little time for A to attend and monitor C’s work. When A then runs into a problem that must be resolved by earlier information, A could either try to interpret the map, where such information should be documented, or directly ask C about the problem and start a negotiation phase instead. Interpreting a documented map is a very time consuming and cognitively difficult task, so it is no surprise that A instead asks C about the current state. In team 1 cooperation and sharing of information becomes a consequence of the sequential constraints, as A is dependent on C’s work and information and therefore continuously monitor C for any information. In team 2 cooperation is triggered by a problem for A. Of course, both teams use both attentive monitoring and negotiation to a high degree, but in different ways. In team 1, attentive monitoring is at the heart of the cooperation. Team members in team 1 mostly share much information because A is monitoring C. In team 2 the team members might not share an understanding of the current state because they process the information themselves and consequently may develop individual situation awareness. When a problem appears they will have to synchronise their 32 representations about the current state first, before discussing future states of the situation. Our study suggest that the serial team generates a hypothesis of the future state of the enemy vs. their units for every message, while the parallel team only generates 1 hypothesis of the 20 messages we have analysed. This dramatic difference may be a consequence of how these two teams share their information. As A in the serial team is being updated continuously by attentive monitoring, they can focus their negotiation time to discuss future and hypothetical states. In the parallel team one member has an elaborated awareness of the situation at the front, and the other does not. Negotiation deals with smoothing out this difference and getting a collective, or symmetrical, awareness of the situation. We propose that the arrangement of information and the accompanying degree of continuous or problem focused cooperation may explain the dramatic differences in hypothesis generation. What we can suggest from the empirical data, is the hypothesis, that a parallel organisation may create more breakdowns between team members, due to different degrees of situation awareness, and that time has to be taken to repair and resolve such differences. Which cooperative arrangement is most successful for military purposes in general is hard to say as it may depend on the pace of the task and the goal. We could not measure any absolute and objective variables of effectiveness in this study due to that the situations the different teams worked with, was dependent and the situations fluctuated. Obviously it is important for any artillery leader to have authorized and well informed decision ground in order to not fire on its own forces or civilians. For this reason a sequentially constrained arrangement may be the best solution. From the theoretical analysis and observations from the empirical data we have generated some hypotheses for future research: • The serial work organization ensures that the participants attend to more of the same information and as a consequence are able to use a common ground or common interpretation as a basis for cooperation, while the parallel organization requires more negotiation and cooperation. • Teams organized for parallel work will formulate fewer collective hypotheses about developments because they share too little information, that is situation awareness, and must concentrate on resolve this difference first. Users of new technology find ways of using the technology that differ from those originally intended by the designers. In this study we have shown that one team uses the technology in an innovative way which makes it possible to speed up the decision process. The effect of the technology in this specific case is related to two conflicting goals; the speed of information flow and the quality of situational awareness of the team and individuals. We form the following hypotheses: • Serial processing is better for high reliability when workload is low due to the availability of physical representation of collective and physical situation awareness. • Parallel processing is better when workload is high and quicker decisions are called for. The field study cannot tell us why the two groups chose different ways organizing the information flow, neither if these different ways of organizing caused their ways of working, communicating or their results. This is the drawback of field studies. However, the field study generated some hypotheses, some of which we investigate in a more controlled study were we compare effectiveness and information sharing practices in serial and parallel organised teams dealing with a dynamic task (see Artman, 1997), other hypotheses may be followed up by other researchers. One issue still unresolved is related to trade-offs. The teams for example have to trade off shared situation assessment against decisions, and they have to trade off filtered information intake versus nonfiltered information. The first may lead to slowness, the second may lead to information overload. Acknowledgements Financial support for this study has been received from the Swedish Work Environment Fund and The Swedish National Defence Establishment. I am grateful to Professor Yvonne Wærn for support in planning the study, help with analysing the data and sorting out the discussion. I would also like to thank E. Hollnagel and (anonymous) reviewers for constructive comments on an earlier draft. Furthermore, I would like to thank all the personnel of the military battalions. References Artman, H. (1997). Co-operation within and between Hierarchical Units in Dynamic Decision Making. Proceedings of CSAPC’97. Baveno, Italy. Artman, H., & Garbis, C., (1998). Situation Awareness as Distributed Cognition. Proceedings of ECCE´98. Limerick Ireland. 34 Artman, H., & Wærn,Y., (1998). Creation and loss of cognitive empathy at an emergency control centre. In Y. Wærn (Ed.), Cooperative process management Cognition and information technology. (London: Taylor & Francis), 69-76 Brehmer, B. (in prep). Distributed Decision Making in Dynamic Environments. Manuscript. Uppsala University, Sweden Brehmer, B. & Svenmarck, P. (1995). Distributed Decision Making in Dynamic Environments: Time Scales and Architecture of Decision Making. I: Caverni, J., BarHillel, M., Barron, F.H., Jungermann, H. (eds.). Contributions to Decision Making (Elsevier Science B.V.), 155-174. Cannon-Bowers J.A., Salas, E. & Converse, S. (1993) Shared Mental Models In Expert Team Decision Making. In N. John Castellan (ed.). Individual and Group Decision Making: Current Issues (LEA: Hillsdale, New Jersey.), 221-245. Duffy, L. (1993) Team Decision-Making Biases: An Information Processing Perspective. In G. Klein, J. Orasanu, R. Calderwood & C.E. Zsambok (Eds.) Decision Making in Action: Models and Methods. (New Jersey: Ablex), 246-361 Endsley, M.R. (1995) Toward a Theory of Situation Awareness in Dynamic Systems. Human factors, 37, 32-65 Flach, J.M. (1995). Situation Awareness: Proceed with caution. Human factors, 37, 149157. Grusenmeyer, C. (1995). Shared functional representation in cooperative tasks. The example of shiftchangeover. International Journal of Human Factors in Manufacturing, 5, 163-176 Heath, C. & Luff, P. (1992) Collaboration and Control. Computer Supported Cooperative Work (CSCW), 1, 69-94. Hutchins, E. (1990). The technology of Team Navigation. In J. Galegher, R.E. Kraut, & C. Egido (eds.), Intellectual Teamwork: Social and technical bases of collaborative work (Hillsdale, NJ: Erlbaum), 191-220. Hutchins, E. (1995). Cognition in the Wild. (MIT press) Klein, G., & Thordsen, M. (1989). Cognitive processes of the team mind Ch28092/89/0000-0046. IEEE. Yellow Springs: Klein Associates. Orasanu, J. & Salas, E. (1993) Team Decision Making in Complex Environments. In G. Klein, J. Orasanu, R. Caldewood & C.E. Zambok (Eds.) Decision Making in Action: Models and Methods (New Jersey: Ablex), 327-345. Salas E., Prince C., Baker P. D. & Shrestha L, (1995), Situation Awereness in Team Performance, Human Factors, 37, 123-136. Salomon, G. (1993). No distribution without individual cognition: a dynamic interational view. In G. Salomon (ed.) Distributed Cognitions - Psychological and Educational Considerations. (Cambridge: Cambridge University Press), 111-139 Samurçay, R. & Rogalski, J. (1993). Cooperative work and decision making in emergency management. Le Travail humain, 56, 53-77 Schmidt, K. (1991). Cooperative work: A conceptual framework. In J. Rasmussen, B.Brehmer, & J. Leplat (eds.), Distributed decision-making: cognitive models for cooperative work (Chichester, UK: Wiley), 75-110. Stammers, R. & Hallam, J. (1985).Task allocation and the balancing of task demands in the multi-man-machine system - som case studies. Applied Ergonomics, 16:4, 251-257. Suchman, L. (1996) Constituting Shared Workplaces. In Y. Engeström & D. Middleton Cognition and Communication at Work (Cambridge University Press), 35-60. Vidulich, M., Dominguez, C., Vogel, E. & McMillan, G. (1994). Situation awareness: Papers and annotated bibliography. Armstrong Laboratory, Crew Systems Directorate. Wright-Patterson AFB OH. AL/CF-TR-1994-0085. Weick, K.E. & Roberts, K.H. (1993) Collective minds in organizations: Heedful Interrelating on flight decks. Administrative Science Quarterly, 38, 357-381. Wellens, A.R. (1993) Group situation awareness and distributed decision making: From military to civilian applications. In N.J. Castellan (ed), Individual and group decision making: Current Issues ( Erlbaum), 267-287. Wertsch, J. (1991) Voices of the mind. (Harvester Wheatsheaf). 36
© Copyright 2025 Paperzz