Dienstag, 22. Januar 2008
Topic and Objectives of the Meeting
Donnerstag, 6. Dezember 2007
Conformity effects in memory as a function of uncertainty and attributional focus
University of Trier, Germany
Previous research has shown that individuals tend to rely on information that is provided by others when they feel uncertain about their own memories. This is because other peoples’ judgments are one source of information about the world “out there” that helps the individual to reduce uncertainty. Based on previous research on memory conformity effects (Walther et al., 2001), the present work investigated the influence of surprise, attributional focus and metacognitive knowledge on individuals’ memory reports. It is assumed that the surprising violation of an expectation often inherent in the observation of a criminal act increases the need to explain the event which in turn enhances social influence. The results of an experimental study indicated that surprise and attributional focus enhanced social influence only if the stimulus was not judged to be particularly memorable.
The Mental Simulation of Other's Action: Concepts, Functions, and Neural Correlates
Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
Recent research reveals humans to mentally simulate actions observed in other people. For instance, perceiving an action triggers a tendency to ‘mimic’ (Chartrand & Bargh, 1999), and activates similar cortical regions as producing that same action (e.g., Buccino et al., 2004). In turn, the premotor cortex has been conceived of as ‘mirror system’, which allows to adapt own actions to a continuously changing environment. However, the psychological mechanisms involved remain open. In our talk, we will highlight the notion of action simulation from a psychological and neuroscience perspective. Second, we aim to explore the mechanisms underlying action simulation. We assume that simulation is (1) a predictive process supported by premotor areas, which furthermore (2) involves semantic action representations. To test our assumptions, participants watched point-light figures performing an action followed by an occluder and a (to be predicted) test posture that was a succession of the action or not. To measure semantic influences, the occluder contained verbs that were congruent or incongruent to the perceived action. If simulation draws on semantic mechanisms, congruent (relative to incongruent) words should facilitate prediction performance. For fMRI-investigations, our paradigm was translated into natural settings. As expected, premotor cortex and posterior perceptual regions appeared as a network for action prediction. The pattern reflects access to the observer’s own motor system when predicting a movement, whereas a memory related task strongly engaged regions for the perception of biological motion. The results will be discussed in the light of disentangling semantic and non-semantic functions in simulating other's behavior.
Who is who? Shared representations of actions and bodies and the consequences for self-recognition
Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
The ability to recognise and to distinguish ourselves from other persons is an essential component of our social behaviour. We normally take it for granted that our body builds an entity independent from other bodies in the outside world and we can easily attribute body states and actions to the self or to another agent. It is the private nature of our bodily sensations which determines a (basic) sense of self. From a scientific point of view however, the ability to recognize one’s own body and actions as such appears problematic and not-well understood. In particular the recent discovery of shared representations of actions, based on mirror neurons, suggests that action attribution is a key computational problem for the sensorimotor system of the brain. The properties of mirror neurons suggest that both self-generated and observed actions activate overlapping neural networks, thus implying a shared representation of self and other at the basic sensorimotor level. How can we attribute actions to the respective agent and more generally, how can we distinguish self and other, given that our brain represents other’s actions in the same way as it represents one’s own? Moreover, if action representations are indeed shared we should frequently make errors in assigning actions to the respective agent. The present paper will give an overview on recent empirical findings showing when misattribution of actions and bodies occur. Furthermore, the possible mechanisms shall be examined that allow us in the end to successfully recognize one’s own actions and ourselves in general.
Group members’ reactions toward defection: the role of shared reality and epistemic motivation.
*University of Rome La Sapienza, Italy
**University of Pittsburgh, USA
***University of Maryland, USA
The present studies, that were stimulated by the theoretical analysis of group reactions to loyalty and disloyalty (Levine & Moreland, 2002), investigated how groups respond to an interesting form of disloyalty, namely defection, which involves leaving one’s current group to join another group. Our three studies (having as participants members of Young Catholic Association, study 1; players of Juvenile Volleyball teams, study 2; and members of Scout groups, study 3) focus on the threat that defectors pose to the group’s sense of shared reality. From this perspective, defectors elicit negative reactions because they jeopardize the group’s sense of shared reality. If this is true, the strength of the group’s negative reaction to defectors should vary positively with the defectors’ threat to shared reality. Furthermore, members with higher Need for Cognitive Closure should perceive more strongly and negatively the threat to shared reality since defection, whatever the defector’s motives represents a change of the status quo and, for high NCC people, a potential threat to order and predictability of the world. Results of all the studies show that: a) high confidence in shared reality appears to reduce the threatening potential of defection episodes; b) high Need for closure appears to induce a more negative reaction toward defection; and c) the relationship between NCC and negative reaction toward defection is moderated by confidence in shared reality: when such a confidence is low high NCC members react more negatively to defection.
Memory conformity in eyewitness situations: Causal inferences and schematic gap-filling errors
Mälardalen University, Sweden
Amina Memon
University of Aberdeen, Scotland
Previous research has shown that individuals who witness and then discuss a crime sequence, can influence each other’s memories of the event (Gabbert, Memon, & Allan, 2003; Gabbert, Memon, & Wright, 2006). It is also known that people may mistakenly infer having seen a cause if they have been presented with the effect of this cause, and that they often make memory errors based on schematic knowledge (Hannigan & Reinitz, 2001). The current study investigates whether conformity effects in eyewitness memory may be moderated by effect presentation and schematic relevance of the witnessed information. In the study, members of a dyad each watch a different video of the same event, a potential theft, where the prime suspect is a man present at the crime scene. Each video version contains unique details seen only by one of the witnesses in the dyad. One member of the dyad actually sees the man steal (the cause), whereas the other witness does not see the theft. Half of the witnesses who don’t see the man steal, see him with the stolen item (effect) and the other half does not. Participants either witness a suspect with a Scandinavian (low schema relevance), or with a southern, Middle eastern appearance (high schema relevance). In one condition, the dyads are encouraged to discuss the event before performing an individual recall test, while in a control condition dyads are not allowed to discuss the event prior to recall. The extent to which witnesses mistakenly remember having seen the theft as a function of activity before recall (discussion vs. no discussion), effect scene presentation, and of the schema relevance of information is examined. Potential mediators of conformity, such as trust in the other witness, and Need to belong will also be investigated.
Cum-fidere. Applying classic research paradigms on social influence to explore the social contruction of confidence concerning true and false memories
University of Rome “La Sapienza”, Italy
In line with researches on qualitative differences between majority and minority influence on memory (e.g. Nemeth et all, 1990) and following the assumptions implied by the etymological Latin root – cum-fidere – of the word confidence, we analysed how different social pressures influence not only individual free recall but also the degree of confidence assigned to memories. We induced a majority pressure using Asch (1951) paradigm (five confederates vs. one naive subject) and a minority pressure, using Moscovici (1976) paradigm (two confederates applying the pressure and three confederates resilient to the pressure vs. one naïve subject). 60 participants were requested to recognize if words presented in a list were included in another, immediately preceding, list. Three couples of lists were created, each couple referring to a different category. Each list included 9 words of a same category (Rosch, 1975). In each couple, the second list was identical to the first one, excluding the crucial item, a word of the same category substituting another word of the first list. The first time, naïve participants answered before confederates, instructed to correctly detect the error (baseline). The second and third time, confederates wrongly recognised crucial items. 15 minutes after the third recognition task, participants received an individual free recall test. Both public recognition tasks and private recall tests were associated with a private judgment of confidence, referred to one’s own responses. Results show that majority pressure induces a higher level of confidence on socially influenced errors, while under minority pressure a higher confidence is assigned to correct remembering.
Attitude Alignment as a Function of Construal
New York University, USA
Research suggests both that attitudes can be remarkably stable across situations, and that they often fluctuate depending on the immediate social context. The present perspective seeks to integrate these findings by using construal level and shared reality theories to inform our understanding of attitudes as functional summaries of socially-shared information. We reasoned that low-level construals of an attitude object should elicit local evaluative summaries, tuned to a particular, shared social context, whereas high-level construals should elicit global summaries, abstracted across multiple contexts. Attitude alignment (at least in temporary social relationships) should therefore occur primarily when the attitude object is construed at a low level. In Study 1, temporal distance was used to manipulate construal. Participants expected to interact with a partner who was favorable or unfavorable toward deporting illegal immigrants, and reported voting preferences for a deportation policy that would be implemented next week or next year. Whereas participants’ attitudes were significantly influenced by partner attitudes for the near-future policy, there was no alignment effect in the distant-future condition. Study 2 zeroed in on our hypothesized process by manipulating construal level directly. Results revealed that participants in a low-level mindset condition aligned their attitudes toward euthanasia with their partner’s attitudes, whereas those in a high-level mindset condition did not. Implications are discussed for the impact of construal-relevant variables (e.g., social distance, linguistic abstraction) on attitude alignment, and a broader theoretical perspective is proposed, in which attitudes are viewed as evaluative summaries that can draw from temporary or ongoing shared realities.
How Communication Goals Determine When Audience Tuning Biases Memory: Shared Reality or Self-Inference Processes?
¹ University of Bielefeld, Germany
² Columbia University, USA
When Belief Sharing is Questioned: Impact of Low Consensus Information on Judgments About Social Groups
Center for Research on Prejudice, Psychology Faculty
Warsaw University, Poland
Perception of belief as shared results in consensual validation (a belief’s experiential status moves from a "subjective feeling" to a “true, objective statement about reality”). Hence, questioning belief’ consensuality should undermine the belief and decrease its impact on social judgment. We hypothesize that this theory holds to the extent that belief is relatively weak, uninvolving, and not central to person’s group identity. When the opposite is true, questioning its consensuality is expected to produce the opposite (paradoxical) effects. Why? In the latter case, low consensus information presumably threatens the conception of the self as a "good” (prototypical) member of the group resulting in activation of whatever strong beliefs accessed by the person in a given moment. We examined how informing participants that their stereotypes are not shared in the group affects social judgment. It appeared that, when judgments were relatively uninvolving and unrelated to participants’ social identity, this treatment weakened stereotypical beliefs on explicit and implicit levels and undermined confidence in their validity (classical effect). However, when we moved to strong stereotypical beliefs (males’ conservative expectations about women’ social roles), the pattern changed substantially. Under group identity primed, low consensus information increased an impact of stereotypical beliefs on judgment (paradoxical effect). Of interest, under personal identity primed, this effect was replaced by the classical effect. This pattern shows that paradoxical increase of a belief strength after the low consensus message is, indeed, specifically related to social identity. In conclusion, questioning belief sharing may produce opposite effects on a momentary intensity of social beliefs. When low consensus information is processed in the cold manner, belief weakening is to be expected. However, when low consensus information is processed in the hot manner (as inconsistent with belief strong and relevant to person’s group identity), belief strengthening seems more likely.
The role of personal common ground in stereotype communication
Université Libre de Bruxelles, Belgium
Anna Clark
Vrije Universiteit Amsterdam
Anthony Lyons
University of Newcastle, UK
The tendency to preferentially communicate stereotype consistent information has been attributed to a motivation to “ground” social information in mutual knowledge (the “common ground”). In line with this assumption, we hypothesize that this SC bias should be stronger when stereotypes are part of the personal common ground (PCG: CG accumulated in the context of specific interpersonal relationship) than when they are not. In Study 1, participants indeed communicated more SC statements and fewer SI statements (about an imaginary footballer) to a friend (with whom they had a PCG) than to a stranger. Study 2 directly manipulated the content of PCG: participants always narrated a story to another unacquainted student. However, before doing so, they either discussed about the main character’s group (i.e., incorporated its stereotype in the PCG) with their audience or not. In order to control for the influence of discussion per se, two control conditions were created in which participants told the story to a student either after discussing the stereotype with another student or without discussing at all. As expected, participants were more likely to communicate SC statements when they communicated with an audience with whom they had discussed the stereotype previously than in any of the three other conditions. Their impression of the main character was also influenced in the same direction, in line with the “saying-is-believing” effect (Higgins & Rholes 1978). The findings of these two studies suggest that a communicational SC bias is more likely to the extent that the stereotype is part of the PCG and that grounding may have different influences on the development of shared impressions as a function of the relationship between communicator and audience.
Mittwoch, 5. Dezember 2007
Speech act, social structuration, and cultural dynamics
Department of Psychology, The University of Melbourne, Australia
Despite social psychology’s ideal of studying social behavior, contemporary social psychology often overlooks the significance of speech act, namely, what we do in saying and its social consequences. In this presentation, I will attempt to provide a theoretical framework in which to consider the relationship of speech act with social structural and cultural dynamical antecedents and consequences. Expanding on Austin, I will discuss a speech act’s representational property (locutionary effect), intended meaning (illocutionary force), and intended and unintended consequences (perlocutionary effect). Although I will touch on what social structural (both social network and group-based social structures) and cultural dynamical antecedents may constrain and enable a certain speech act, the central focus will be placed on a discussion about the interactive grounding process in the generation of a speech act, and potential socio-cultural consequences of speech acts in the production and reproduction of social network and group-based social structures and culture as dynamically evolving distributed collective representations. In particular, I will characterize grounding as socially coordinated processes in which interactants align their collective identities and collective representations, and then argue that social structuration and cultural dynamics – formation, maintenance and transformation of a social structure and collective representations – are inseparable dual consequences of grounding, and that the human capacity to form and sustain a large-scale collective is inexorably linked to our capacity to construct shared representations about the collective itself.
The role of expertise and narratorship in the creation of a shared memory
New School for Social Research, New York
Conversations can transform initially distinctive memories of a shared experience into shared memories, in part because one conversational participant can impose his or her version of the past on other participants. The dynamics underlying the conversational implantation of memories are poorly understood. Two factors that have been mentioned as possible mediators through which conversations promote the formation of shared memories are narratorship and expertise. These factors, however, are confounded: Is an individual more likely to impose misleading information because she is a perceived expert or because she talks a lot? We will discuss experiments that piece about the role of narratorship and expertise in the formation of shared memories through conversation. We find expertise alone does not lead to a significant advantage in the creation of shared memories. It is through the dominant role an expert might play in a conversation that he or she can reshape subsequent remembering.
The Role of Uncertainty for Audience-Tuning Effects on Memory
1 University of Bielefeld, Germany
2 University of Aberdeen, Scotland
Communicators often modify their messages about a topic toward their addressee’s attitude on this issue. This may lead to subsequent biased retellings of the original events. These audience-tuning effects could often be replicated over the last three decades with the Communication Game (e.g., Higgins, 1992). However, so far psychological researchers have focused on verbal material used to describe a target person. In this study we tried to investigate audience-tuning effects on memory in a more ecologically valid context. For the first time, participants watched a video that depicted ambiguous behaviours of a target person (instead of reading an essay). We were able to replicate audience-tuning effects for both memory and impression: Participants liked the target person more and remembered him in a positively biased way, after tuning a message to an audience who liked (vs. disliked) the target person and vice versa. Importantly, this was only the case after participants received a negative (vs. positive) feedback regarding their capability to form a reliable impression about other people. In sum, these results support the notion that high uncertainty about the communication topic is a necessary precondition for audiencetuning effects on memory (Echterhoff, Higgins, Kopietz & Groll, in press).
Effects of social verification (but not of collaboration) on memory and social coordination
Columbia University, USA
It has long been an axiom in social psychology that individuals rely on other people for their understanding of the world (e.g., Sherif, 1935; Festinger, 1950; Hardin & Higgins, 1996). The “Communication Game” (Higgins & Rholes, 1978) is an experimental paradigm that has been used to demonstrate that social verification can yield lingering, shared memories (e.g., Echterhoff, Higgins, & Groll, 2005). We hypothesized that the memory effects of receiving social verification from another person go beyond the effects of mere collaboration. Moreover, we hypothesized that such socially constructed memories can mediate behavioral effects like social coordination and trusting behavior. In addition to basic social-verification conditions, separate conditions were administered where participants were able to collaborate with a team member without a chance to verify each others’ views. The successful social verification group showed significantly stronger memory effects (saying-is-believing tendency) than any of the other three groups: failed social verification, successful mere collaboration, and failed mere collaboration. In an unrelated-studies paradigm, participants in the successful social verification group later demonstrated significantly more trust toward the other person and three other individuals in a coordination situation with actual monetary payoffs (weak link game) than did any other of the groups. We discuss the implications of the results for viewing shared memory establishment as a basic mechanism for social coordination, and consider components in a trust taxonomy.
Shared Reality in Communication
University of Bielefeld, Germany
E. Tory Higgins
Columbia University, USA
A key topic of social psychology is the human tendency to share inner states, such as beliefs, judgments, or, more generally, representations of the world. The power of social sharing derives from its potential to satisfy fundamental motives, in particular relational needs (feeling connected with others) and epistemic needs (achieving a reliable understanding of the world). Shared reality emerges as the product of the motivated sharing of inner states. The talk focuses on communication as a pathway of creating a shared reality and on the epistemic functions of sharing. Research investigating communication effects in the saying-is-believing paradigm has shown that tuning messages to the audience’s attitude can shape the communicators’ own representations of and memory for the message referent (here, a target person). We review a body of recent studies demonstrating that this effect occurs to the extent that communicators achieve a sense of shared reality about the referent with their audience. The evidence suggests that the effect depends on whether communicators are motivated to create a shared reality through audience tuning, whether they accept the audience as an appropriate co-constructor of reality, whether they experience their communication as successful, and whether they connect to the audience’s inner state about the referent. The impact of manipulations of shared reality is typically found to be mediated by the communicators’ epistemic trust in the audience and their audience-congruent message. We discuss pathways of social sharing other than communication and differences between shared reality and related concepts.
Flashbulb memories as shared memories: Conceptual and empirical issues
1University of Bari, Italy
2University of Louvain, Louvain-la-Neuve, Belgium
Flashbulb memories (FBMs) are defined as vivid, detailed, and long-lasting memories for attributes of the reception context of relevant public news (Brown & Kulik, 1977). Two sets of factors have been considered to account for their formation. The first encompasses the so-called encoding factors, corresponding to the characteristics of the emotional experience associated with learning of the news, and its cognitive antecedents (Finkenauer et al., 1998; Pillemer, 1984). The other set involves social processes both preceding and following the original encoding. When a given event takes place, expectations, knowledge and attitudes are spread out in social contexts about the fact and its protagonists, and this influences the way individuals perceive and remember their experience (Curci, Luminet, Finkenauer & Gisle, 2001; Conway et al., 1994). Additionally, the emotion felt when learning of the news triggers social sharing processes, which develop both in private conversations, and through the mass media (Finkenauer et al., 1998; McCloskey, Wible, & Cohen, 1988; Wright, 1993). The present paper aims to provide both conceptual and empirical basis for considering FBMs as a special class of memories deeply determined by social factors. Three correlational studies assessed the impact of emotional and social processes, by focusing on both the measurement of the construct and the structural process of formation. Findings from the three studies jointly considered showed that FBMs are highly influenced by social factors, thus representing a significant connection between individual and social dimensions of memory. Implications for a social account on autobiographical memory are discussed.
Cognitive Accessibility of the O.J. Simpson Trial Regulates Interpersonal, Inter-Ethnic Relationships
University of Missouri, USA
Curtis. D. Hardin
Brooklyn College, USA
Shared reality theory postulates that interpersonal relationships are regulated by the degree to which participants share experiences and attitudes (Hardin & Conley, 2001; Hardin & Higgins, 1996). We were interested in whether these shared beliefs can promote positive interpersonal relationships. This postulate was tested by examining the consequences of activating thought about the O.J. Simpson murder trial on actual interactions among Blacks and Whites at three time points: directly following the trial, five years after the trial, and ten years after the trial. In three experiments, mixed-ethnicity and same-ethnicity dyads participated in cooperative problem-solving tasks after being reminded of the Simpson trial. Because the trial represents a dimension of belief discrepancy between Blacks and Whites but a dimension of shared beliefs among Whites and among Blacks, we hypothesized that activating the trial would affect interpersonal interactions differently depending upon the ethnic composition of participant pairs. As predicted, activating the Simpson trial caused the quality of interpersonal perceptions and behaviors to decrease for mixed-ethnicity dyads, but increase for same-ethnicity dyads. The latter finding concerning the increased positive perceptions among same ethnicity dyads is uniquely predicted by shared reality. These findings indicate that shared reality promotes the development of positive relationships. The findings are strongly consistent with shared reality theory but only weakly, if at all, supportive of other alternatives, such as social identity theory, a similarity-breeds-liking perspective, or a stereotyping perspective.
Situating construal
Free University Amsterdam
Events can be construed in different ways. Level of construal can vary from concrete, focusing on the local and specific aspects, to abstract, which emphasizes the general or global features. Research to date shows various influences on construal, e.g., the psychological distance (i.e., temporal or spatial) of the event or object, or motivational and cognitive processes such as those underlying the linguistic intergroup bias. The question is what determines construal in any given context? By situating construal in a social context we show how construal level can be predicted based on the relative relevancies of concreteness and abstractness. In our research we have examined construal in a conversational context. Communicators were asked to talk about an event to another person. The temporal distance of the event in question (close vs. distant) and the conversational partners’ shared knowledge of the event present, absent, unspecified) were manipulated orthogonally to create different conversational contexts for talking about the event. It was found that temporal distance information directed construal level when no information about the conversational partners’ existing knowledge was given (unspecified condition); close events were construed more concretely than distant events. When information about the conversational partners’ shared knowledge was present, however, this determined construal level regardless of the temporal distance of the event; construal was more concrete when shared knowledge was present. These results suggest that construal level is flexible and shifts as a function of the conversational contexts as different cues have different implications for the relevance of concreteness and abstractness. In situating construal and examining the interpersonal and social processes our work adds to existing research that has examined construal as the product of a learned association (e.g., construal level theory) and cognitive and motivational processes (e.g., the linguistic intergroup bias).
Audience Tuning Effect on Memory and Subsequent Communication: The Role of Shared Reality and Perceived Consensus
Boyka Bratanova & Yoshi Kashima
Department of Psychology, The University of Melbourne, Australia
The current research used the Saying-is-Believing (SIB) paradigm with a group as a target of communication and included subsequent communicative acts to examine stereotype formation resulting from self-generated communicative influence. A series of four experiments supported the existence of the SIB effect. It was moderated by the experience of shared reality when recall was administered after the first message production or after one week delay. When recall was administered following communication to a second audience unacquainted with the target group, however, the SIB was negatively moderated by shared reality (i.e. communicators who scored high on shared reality did not exhibit a memory bias, while those who scored low on shared reality showed biased recollections). It seems that having to inform a newcomer poses external pressures to communicators, which reverses the effect of shared reality. A meta-analysis across the experiments also showed that messages produced for subsequent audiences conveyed the bias from previous messages independently from the experience of shared reality – a phenomenon we called Saying –is – Diffusing (SID) effect. Apparently, communicators do not need to endorse the biased information in order to further transmit it to their in-group members. Nevertheless, the bias transmission in communication to an audience acquainted with the target was moderated by the perceived consensus about the target group characteristics. This finding suggests that contrary to memory, where genuine endorsement is required for the recall bias to occur, communication is guided by more exogenous concerns like matching the already existing stereotype in one’s in-group.
Social memory: individual, collective or social influence? An integrative working model
Hartmut Blank
University of Portsmouth, UK
The research domain of social memory contains three major subfields, research on (1) social cognition, (2) collective memory and (3) social influence on memory. In the first area, “social memory” is conceived as individual memory with social contents (e.g. person memory). The second area focuses on properties and contents of memories held at the level of groups or societies, often presupposing that these differ from individual memories (if the possibility of individual memories is not denied altogether). The third area covers changes in individual memory as a function of the persuasive influence of one other person (most of eyewitness suggestibility research can be construed to fall in this category) or of group pressure. I will argue that social influence on memory constitutes the essence of social memory and also can function as a theoretical link between individual and collective memory. I will present a working model of remembering that integrates perspectives of (individual) memory psychology and social psychology. This model describes remembering as being more than just retrieving information from memory. It also includes memory conversion (Tulving, 1983), the adaptation of memory information to the purposes of a memory task. Classic memory retrieval constitutes only the first stage in this process. The second involves the formation of memory beliefs, which may be converted to memory statements in a third stage. At stages 2 and 3, various forms of social influence can take place, which can be roughly group into informational (at stage 2) and normative (at stage 3) influence.
How shared representations are created, updated and used in the coordination of task-related conversation
Adrian Bangerter
Institut de Psychologie du Travail et des Organisations University of Neuchâtel, Switzerland
Shared representations emerge from joint dialogical projects, which they serve to coordinate. This contribution examines how discourse markers and acknowledgment tokens (okay, all right, uh-huh, m-hm, yeah) are used to coordinate task-oriented dialogue. These words have previously been studied as devices for managing turntaking. In contrast, the theoretical perspective adopted here conceptualizes language use as a form of joint activity which must be coordinated by participants (Clark, 1996). Acknowledgment tokens and discourse markers coordinate transitions in discourse. In task-oriented dialogue, participants must move from one part of the task to the next. They need to create and update a shared representation of where they are in the task. They use dialogue to coordinate two kinds of transitions: vertical transitions, or entering and exiting joint projects; and horizontal transitions, or continuing within joint projects. I review evidence from several types of tasks that uhhuh,m-hm and yeah are used for horizontal transitions, and okay and all right for vertical transitions. These words form a conventional system of contrasts specialized for navigating joint activities. For the small group meeting, this work has implications for understanding the conventional nature of shared representations and how they serve joint activity.