Primacy of Action: An Advanced Interdisciplinary Workshop
Ashburn Hall, Manchester
13-15th December 1993

The workshop on the study of action was the second to take place in Manchester. It was attended by researchers in the fields of psychology, sociology, discourse analysis, archaeology, anthropology, and computer science. The reason for organizing the workshop was to develop a coherent ecological approach to situated action through reference to a wide range of concrete examples, and, in the process, combine the resources of the different disciplines. The workshop brought together researchers for whom the concept of activity is the primary theoretical construct. Special emphasis was put on the following issues

The workshop was organized by


ABSTRACTS

Completions in three-part sequences
Charles Antaki, Felix Diaz and Alan F Collins
Psychology Department, University of Lancaster

As ordinary language users we have an intuitive notion of a speaker 'completing' a previous speaker's utterance. The straightforward account of such completions is that they are purely syntactic: that they work by offering a syntactically coherent termination of the previous utterance. This syntactic cohesion is, we argue, only part of the story. We want to show that completions are evaluated for their footing. That is to say, we argue that completions normally work as three-part sequences, in which a syntactic completion is offered in the second turn and then accepted or rejected in the third turn. The original speaker accepts or rejects not merely the propositional content of the putative completion, but also the footing on which it is uttered (what Levinson calls its participant status). Most completions in the corpus we analyse stay within the typical case of the 'author', but there are significant examples of other statuses. We establish the case for using participant status as a criterion of successful completion, list some of its features, and speculate on variations in how it might be maintained and challenged.

Towards an Ecology of Theorising: The Role of Intentionality in Physics
Ivan Blair
Dungannon, Northern Ireland

Physics is often perceived as a scientific domain where theory is objective and observerneutral. I claim that this view of physics is inaccurate. Physicists have no choice but to take decisions and make stipulations in order to get both experiments and theories off the ground. Such is the fundamental nature of the intentional structure at the base of physics. This claim win be illustrated by a brief consideration of the following three cases: measurement (the identification of observables and the choice of a suitable standard unit); the application of a theory in a real situation (inertial frame selection in Newtonian mechanics); and construction of a theory's Founds of intelligibility (stipulation of the equivalence of gravitation and acceleration in general relativity). These results, however, do not entail a lapse into subjectivism and relativism. The lesson to be drawn is, rather, that even in physics there can be no de-coupling of theorist, theoretical framework and domain of investigation. Ecological psychology has drawn attention to the need to consider the organism in its environment, rather than as a black box subjected to stimuli. To this we must add that the theorist, or scientist, too, must have a place in the account, a place which is determined by the intentional structure of the account. There is an inescapable reflexivity which becomes evident when we consider the case where the organism in the account is the theorist. Consideration of this case from the perspectives of both the foundations of physics and ecological psychology should be mutually illuminating.

Talking through Design: Requirements and Resistance in Cooperative Prototyping.
John Bowers and James Pycock
Department of Psychology, University of Manchester

Some analyses are presented of talk between designers and a potential user in a participatory design session where a prototype application was worked with to determine future requirements. We explore the ways in which design suggestions are formulated and argued for, and how requirements emerge as a negotiated product of interaction. On this basis, we urge a reappraisal of user participation in de sign and of the relationship between prototyping and user-requirements. We conclude by offering a notion (gradients of resistance in design space) to help understand the interplay of the social and the technical in design.

On acting in order to see things better.
Roddy Cowie
Psychology Department, Queen's University, Belfast

Vision is active in the straightforward sense that we may use our ability to act in order to improve our view. Gibson pointed this out long ago, but the point has not been developed much. We have carried out a series of studies on the issue and explored their implications for the geometric foundations of vision. We have shown that perceivers are better at judging objects' geometry when they are controlling their viewpoint than when they receive the same sequence of views passively. Using computerised equivalents of the same stimuli lets us show which views (and sequences of views) observers seek out, and suggest why they are particularly useful. These findings match results from a converse type of study where subjects' views are limited to certain sectors of space. A non-Gibsonian, but reasonable theme emerges throughout the research: controlling viewpoint depends on awareness of the geometric regularities that are likely to occur in the world. The studies point to a way of understanding perception through motion which diverges from standard computational analyses. It builds into the analysis of perception the fact that the world is not a random place, partly because our actions are part of it, and we control them. This is not dramatic in philosophical terms, but it is interesting in the sense that the ideas suggest specific mathematical and computational developments.

Collective formulation: some instances from a University society's meetings
Felix Diaz
Department of Psychology, University of Lancaster

'Collective formulation' is a conversational process through which several speakers make a formulation in a collective voice. The construction and ratification of a formulation as collective by participants produces a collective voice from a series of individual utterances. I discuss examples from the meetings of a University society, applying an ethnographically-informed conversation analysis. I focus on conversational resources including the mutual completion of utterances, the linguistic reference to co- participants in the first person plural, and the contraposition of arguments in two-part structures. I argue that collective formulation should be understood by reference to the participants' construction, recognition and use of collective entities in formulation sequences.

An eco-structural framework for investigating social-cognitive abilities in children.
Michael A. Forrester
Department of Psychology, University of Kent

This paper outlines a conceptual framework for investigating and understanding young children's active engagement and participation with the social world. This approach takes up a number of ideas from within and beyond psychology, notably from conversational analysis, micro-sociology, pragmatics and from the ecological perspective, the notion of affordances. The eco-structural framework is an attempt at formalising the active and participatory nature of children's social cognitive development. It also seeks to articulate how the dynamics of participation can be studied, particularly through employing the affordance metaphor. In doing so it is suggested that conceptions of representation require re-formulation, emphasising narratological orientations and how and these bear upon conversational processes. This framework should allow us to examine the correspondence between learning to display intentionality and the requirements of participant 'structuration' of conversational affordances. The discussion will centre upon two aspects of the framework: the ecolocical perspective and conversational analysis. Six propositions will help form the basis of discussion, (four with reference to the ecological positions (EP), two with conversational analysis (CA)) :

(EP) proposition 1: predominant orientation of sensory-cognitive processes (that is, arising from, but not exclusively, visual perception), leads to our engaging in constructivist conversational practices which build upon our skills, or predispositions, to detect and extract affordances.

(EP) proposition 2: In conversations, and in the construction of them, we make available and use patterns and structures (of talk/language) so as to signal and identify those aspects of the on-going talk which have to be picked up, ignored, made recognisable, or whatever.

(EP) proposition 3: conversational affordances are socially constructed representational constructs - directly recognisable as structurally invariant components of conversations (utilised and made available by participants). Representations are both recognised and provided by participants as affordance structures in and through talk.

(EP) proposition 4: Affordances are social- cultural constructions which bear upon conversational/representational process only as an indirect result of evolutionary processes (e.g. as recognitory predispositions)

(CA) proposition 5: structural patterns identified by the conversational analysts could be considered from an ecological viewpoint, i.e. as affordances and affordance structures.

(CA) proposition 6: structuration describes the observation that participants themselves utilise the affordance structures in talk such that they both recognise their occurrence, produce them and employ an orientation to their occurrence and on-going development during participation in conversation

Summary comments will consider the insights provided by Goffman's micro-sociological analysis of participation contexts.

The Doors of Perception
Paul Graves
Department of Archaeology, University of Southampton

This paper develops an account of the history and typology of entrances and exits. I suggest that doors of all kinds form a nexus for joint action. But in order to do so they must conform to traditions and conventions, as we as to more 'natural' ecological and functional principles. By examining the ways in which such conventions work, or indeed don't work, I argue that material culture plays a key role in scaffolding social action.

Actions and Biological Motion: A method for generating dynamic displays in experimental settings
Paul Hemeren
Cognitive Science, University of Lund

The patch-light technique, as developed by Johansson (1973,1975), demonstrated the effective use of kinematic information for the perception of biologically motion based actions, i.e., walking, running, riding a bike, dancing, etc. By placing small lights on the joints of a human body and filming various action sequences against a dark background, naive subjects were able to recognize the various sequences on the basis of the resulting flow pattern of the lights. It is seldom the case, however, that people actually see other people or objects emitting light. Objects in the optic array typically reflect light.

Runeson & Frykholm (1981,1983) employed a slightly different method for producing the patch-light displays. Instead of lights, they used reflector material on the joints of a person. The person was then filmed with a video camera against a dark background producing a phenomenon similar to the Johansson displays. Runeson & Frykholm (1983) demonstrated a number of effects using this technique. The principle of Kinematic Specification of Dynamics (KSD) was used to explain such effects as the perception of throwing length, the weight of a lifted box, gender recognition and deceptive intentions. The KSD principle roughly states that the kinematic patterns given in the optic array of the flow patterns of the reflected light is sufficient to specify the dynamics inherent in human movement and interaction with objects. That is, kinematics contain information about the dynamics of movement.

Given the patch-light technique for displaying biologically based actions and the findings by Runeson & Fyrkholm, a further step towards investigating the perception and recognition of actions is to develop a more flexible and accessible method for generating dynamic displays which could be shown in real-time on a computer screen. One potential drawback with the video tape format used by Runeson & Frykholm, for example, is that the tape quality is not held constant across subjects. Another limitation is the difficulty of editing movement sequences. As an example, one might wish to investigate the effect of deleting certain patches that may convey more information than others. With computer based displays, this presents no difficulty, and the motion can be kept constant while deleting or adding patches. No re-filming of a sequence is necessary. The displays can also be shown in a number of different orientations while keeping the motion constant.

Using a technique similar to Runeson & Frykholm, I have generated a number of displays (about 50) by simply digitizing the video taped images frame-by-frame and then reconstructing the sequences in an animation program. Once the sequences were stored on the computer, they were used in a reaction time study for action identification. We have also developed a general purpose experiment program to run the displays and to accept a variety of subject responses. The method of illuminating an actor with the reflector material around the joints not only produces reflected light but also produces images that contain directional illumination. That is, the amount of light reflected from the material is different depending on where the source of light is coming from and from the angle of filming, in the case where the light source and the camera are at different positions.

Another aspect of using this method is its potential for investigating the information conveyed by surface structures. The surface information in the displays is available through factors such as occlusion and interaction with objects. In relation to KSD, kinematic patterns can also relate spatial information to an observer as well as information about the dynamics of human movement and interaction.

Perceiving and acting as value-realizing activities.
Bert H. Hodges
c/o/ Department of Psychology, University of Southampton; also of Gordon College, Wenham, MA, USA

My main concern has been to explore the claim that ecological activities, exploratory and performatory, are value-realizing (e.g., Hodges & Baron, 1992), not just lawful or rule- following. Ecological and social constructionist theorists (cf. Harre & Secord, 1972; Turvey, 1992) have noted that both lawful constraints and rule-following, goal-seeking circumstances must be "added" or "blended" to yield accounts of coordination (e.g., walking; Turvey, 1990) and "prospective control" (e.g., a bird changing its flying so as to land safely on a branch; Turvey, 1992). Our claim is that this "enigmatic" blending/adding process is guided by values-- global, ontological constraints on the constitution and dynamics of an ecosystem that guide the development of competent cognition and appropriate action. This value-realizing approach will be used to explore the possibilities and problems of ecological research on human affordances and accuracy (e.g., Heft, 1993), as well as the social psychological research on pragmatics and accuracy (e.g., Fiske, 1993). It appears that many studies have been unecological, restricting the both the possibilities and consequences (Heft, 1989) inherent in affordances. For example, experimenters have selected the affordances (e.g., step-on-ability) and the constrained effectivities (e.g., bodily orientation, intention) in such a way that the value-realizing character of ordinary perceiving and acting have been occluded. The tasks have often required analytical judgments rather than effective action (cf. Heft, 1993) and have ignored the social and moral context that ordinarily motivates and judges our activities (cf. Pufall & Dunbar, 1992).

Fiske, S. (1993). Social cognition and social perception. Annual Review of Psychology, 44.
Harre, R., & Secord, P. (1972). The explanation of social behaviour. Oxford: Basil Blackwell.
Heft, H. (1989). Affordances and the body: An intentional analysis of Gibson's ecological approach to visual perception. J. for the Theory of Social Behaviour, 19, 1-30.
Heft, H. (1993). A methodological note on overestimates of reaching distance: Distinguishing between perceptual and analytical judgments. Ecological Psychology, 5, 255-271.
Hodges, B. H., & Baron, R. M. (1992). Values as constraints on affordances: Perceiving and acting properly. Journal for the Theory of Social Behaviour, 22, 263-294.
Pufall, P. B., & Dunbar, C. (1992). Perceiving whether or not the world affords stepping onto and over: A developmental study. Ecological Psychology, 4, 17-38.
Turvey, M. T. (1990). Coordination. American Psychologist, 45, 938-953.
Turvey, M. T. (1992). Affordances and prospective control: An outline of ontology. Ecological Psychology, 4, 173-188.

Paperwork
John Hughes
Department of Sociology, University of Lancaster

No abstract available

Enculturation versus Enskillment
Tim Ingold
Social Anthropology, University of Manchester

Orthodox models of cultural learning rest on the premiss that culture exists as a context-free body of information, available for transmission outside the contexts of its practical application. I argue that this premiss is untenable, suggesting instead that what each generation contributes to the next are not rules or schemata for the production of appropriate behaviour, but the specific conditions of development under which successors, growing up in a social environment, acquire their own embodied skills and dispositions. Thus learning is a matter of enskillment rather than enculturation. This conclusion, however, has radical implications for the way we think about the relations between biological and cultural variation. Instead of supposing that the human being comes into the world innately pre-equipped with mechanisms for the acquisition of cultural information, we need to recognise that the differences we call cultural are themselves biological, established in the human organism through a process of development. This recognition, however, calls for a restructuring not only of the psychological theory of learning but also of the biological theory of genetic inheritance.

Work activities as basic units of analysis for CSCW research and design.
Kari Kuutti
Department of Information, University of Oulu, Finland

CSCW (Computer-Supported Cooperative Work) is a new research field emerged during the 1980s as a result of several strands of critiques against the dominating practices in information system design. One characteristic feature of the field has been various attempts to utilize knowledge drawn from different branches of social sciences in system design. The progress in cumulation of knowledge has been slow, however, and it is not an exaggeration to say that fragmentedness and confusion are prevailing. We can say that the field is suffering from the lack of commonly recognised unit of analysis, an entity where different threads of research could become conveniently connected.

What are the characterizing features of the "CSCW terrain" to be studied and designed? Obviously the terrain is information technology artifacts used in the work by active subjects belonging to a working community and culture - and perhaps other communities and cultures as well - and as a whole embedded in a larger socio-economical and political production system. Of course there is nothing particularly new in any of the parts brought together. It is the complexity of affairs when all these parts are brought together and the attempt to deal seriously with them which is distinguishing CSCW from older approaches to system design. We should be able at the same time handle individual and societal - perhaps even several levels of societal - issues and change processes, and technological artifacts which penetrate all individual and societal relations, shape them and become shaped in the same process. And this is not an easy task by any means. It is calling forth a unit of analysis where knowledge from multiple disciplines can be combined.

The paper suggests that the concept of activity from the Soviet Union-originated research tradition called Activity Theory might be useful in delineating the basic unit of analysis for the CSCW research.

Studies of Action in Ethnomethodology, Activity Theory and Cognitive Science
Ivan Leudar
Psychology Department, The University of Manchester

(abstract not available)

The activity theory and the ecological approach to visual perception
A. Logvinenko
School of Psychology, The Queen's University of Belfast

The well-known Russian psychologist A. N. Leont'ev (1903--1979) put forward an original approach to visual perception in the context of what he called activity theory. It shares a few crucial features with Gibsonian approach to visual perception. Both theories suggest that what is to be perceived is not the physical world but the environment. In the context of activity theory the latter is understood as a result of externalizing the inner forms of an animal's behaviour. Particularly it implies that an animal perceives something from the external world insofar as it is potentially involved in the animal's behaviour. Thus, both theories assume that the very act of perceiving is an overt action implying some motor activity. More properly, perception is assumed to be an aspect of behaviour. The operational structure of perception has been studied extensively under both approach.

The dynamic modelling of activity with ActionSimulator
Andrew Monk
Human-Computer Interaction Group, Department of Psychology, University of York

The effective design of equipment and procedures for process control and human- computer interaction depends on a clear understanding of the environmental and intentional constraints on the actions available to operators. One way of achieving such an understanding is to build an abstract model that can be incorporated into the specification of the system.

There have been a number of attempts to use abstract models in the design of interactive computer systems. These typically fall into two groups: (i) task-action mappings and (ii) action-effect mappings. The former are "cognitivist" planning models that describe how some goal is mapped down to actions with the system (for example Payne and Green's TAG and Card Moran & Newell's GOMS). The latter are system models describing how the actions of a user effect the system (for example Dix and Runciman's PIE model). Neither account pays any attention to the influence of the perceived environment on the action taken nor the cyclic nature of activity whereby action changes the environment and thus changes the opportunities for further actions.

ActionSimulator allows a designer to build simple production systems that model the high level behaviour of a human-computer system. Its key characteristic are:

My presentation will consist of a demonstration of ActionSimulator. This demonstration revolves around the example of a group of operators controlling a nuclear power station (taken from Hollnagel 1993). The components of the model are: (i) actions - these are events initiated by the operators or coming from the plant that may change the state of the model, e.g., "Send mechanic to AFW", "AFW starts"; (ii) states in the form of world states (e.g., "Cooling OK" and "AFW working") and history (e.g., "Mechanic dispatched") and (iii) rules - each rule has an action as its subject. It specifies the states that are pre-conditions for that action changing the states in some way.

The demonstration will show how an ActionSimulator model can be used to characterise what knowledge (world states and history) is required by the team member responsible for a particular action. Also how it can be used to work out the consequences if this knowledge is not available e.g., if one team member does not know that the rest of the group has decided that "AFW broken" is true.

ActionSimulator consists of an Excel spreadsheet and associated macros and is publicly available via anonymous FTP.

Referential Communication ("Naming") as Coordinated Achievement
William Noble
Psychology Dept, University of New England, Australia

An ecological approach to coordinated activity among organisms must be concerned to relate the communicative behaviour of those organisms to prevailing conditions within the physical environment. Coordination among organisms depends on communication, irrespective of whether the coordinated activity is cooperative or agonistic. Arguments have been raised about a role for intentionality in describing behaviour as communicative, but that matter can be left unresolved. Intentionality implies self- consciousness about goals, and hence relies on communication that is referential (applies names to things). An organism's communicative behaviour, and its effects, can be observed, not simply responded to (by self-consciously aware humans); and its significance can be witnessed in virtue of human observation of its effects.

The bulk of animal communication, be it in the form of visible, audible or chemical signs, has the effect of drawing attention to the communicator and/or warding off the other. Some communicative signs have the effect of making another organism aware of both the communicator and of something in the environment besides the communicator. The vocal signs of vervet monkeys can be described this way. In their case the coordination achieved takes the form of combined visual attention to various sources of external threat. That is a significant behaviour. It does not seem to be referential; but it underpins a subsequent capacity for reference to environmental events and objects that emerges in evolutionary terms. In the case of the monkeys, the signs which achieve coordination are not, themselves, objects of the monkeys' attention. A context would have to arise in which coordinated activity could induce combined attention to signs themselves. Referential communication results when coordinated activity takes the form of combined attention to external objects in association with signs which are also a combined (and "externalised") object of attention.

Reflexive research: The context of action
Ian Parker
Department of Psychology and Speech Pathology, Manchester Metropolitan University

This paper is concerned with reflexivity in research, and the way research is grounded in the operations of the psy-complex in social psychology. It discusses the reflexive engagement of the researcher with data, and the construction of the identity of the researcher with reference to professional bodies. An analysis of a document produced by the British Psychological Society is presented to illustrate conceptual issues addressed in the first sections of the paper. A central argument is that qualitative research in general, and a focus on reflexivity in particular, requires theoretical grounding. In this case it is argued that such theoretical grounding can usefully draw on developments in discourse analytic, deconstructionist and psychoanalytic social research. The opposition between objectivity and subjectivity is deconstructed, and psychoanalytic conceptual reference points for an understanding of discursive subject positions are explored with particular reference to the location of the researcher in the psy-complex.

Perception of Object Through Their Dynamics: Some Experimental Evidence
Marina A. Pavlova
Institute of Psychology, Russian Academy of Sciences, Moscow, Russia.

Recent developments in the study of the visual information afforded by dynamic events are described. We investigated object perception through motions of its elements (using biological motion patterns). It was found that: a) the 3- year-old children are able to recognize the biological motion patterns, and that the performance of 5-year-old children is as high as that of adults; b) stimulus inversion (180 deg) completely prevents its spontaneous recognition, and a priori information about the object its orientation and additional background elements cannot mediate its correct perception; c) Observers are equally able to recover object structure through motion in spite of the nature of transformation used (usual/reverse), but stimulus interpretations based on usual transformation are more stable. These findings suggest that the perception of an object through motion depends not only on relative motions of its elements, but on some ecological constraints in visual perception. Emotion attribution as a function of perceived dynamics of abstract figure was investigated: a) perceived unstability of the figure is related to some negative emotions ascribed to it; b) the strength of tense positive emotions attributed are inversely correlated with their physical deviation from vertical. Similar evidence was yielded by a preliminary study of preschoolers aged 5-6 years. The findings are discussed in the context of an ecological and system approach to visual perception, where the concept of affordance is considered as a kind of system quality.

Action and Attribution
John Pickering
Department of Psychology, University of Warwick.

Action is the essence of psychological being. FFrom Brentano to Harre, there is a thread of psychological and philosophical thinking which takes the act as its starting point. But this thread is obscured in cognitive science by the functionalist emphasis on cognitive precursors to action.

Attributions of intentionality in the folk psychology sense are made on the basis of mental dispositions like "being purposive", "intending", "having the will to do something", "having a goal", "seeking an end" and so on. Cognitive science reduces folk attributions like these to cognitive structures that are supposed to exist within the actor as the necessary precursors of action. These represent things such as goal states, present states of the world, possible actions and their consequences and so on. Brentano's epistemic sense of intentionality, the relationship of 'aboutness', is substituted for the dispositional folk psychology sense. Internal cognitive structures are said to posses the property of being 'about' external things. This property allows actions to be planned and executed.

But, as the frame problem and the symbol grounding problem reveal, this is unlikely to be the basis for much, if not most, natural action. With respect to natural action, cognitive structures may be secondary rather than primary. Dreyfus makes something like this case when he points out that action can occur without cognitive precursors. As he puts it: "to act in a domain does not require a theory of that domain".

This position can be developed in the light of Brooks' work on action. Brooks provides an explicit and practical account of how apparently purposive action can emerge when an appropriately structured robot engages with the environment in which it is designed to act. Note 'structured' and not 'programmed' - there are no representations of actions nor of the environment within the control system of Brooks' robots. What produces purposive action is the presentation, not the representation of the environment.

Although Brooks' research deals with relatively simplified situations, it nevertheless useful for considering where intentionality may be attributed in the vastly more complex arena of human action. The suggestion will be made that internal attribution, both in folk psychology and in cognitive science, is given undue weight. Agency, the production and shaping of action, resides in the situation-actor system, not in the actor alone."

Coordinating Others' Knowledge in Infancy
Vasudevi Reddy
Dept. of Psychology, University of Portsmouth

Traditional approaches to understanding others' knowledge seem to assume that it is a more or less static proposition or set of propositions in the head' and therefore difficult to grasp. Studies by developmental psychologists have shown that when asked to report on such propositions held by another person, children do not reveal an understanding of knowledge and ignorance in others until about 4 years of age. The present paper reports the findings of a naturalistic study of an 18 to 24 month-old infant's declarative acts - Tselling' and Tshowing' things - to other people in everyday interaction.

The results show spontaneous discriminations between knowing and unknowing others made by the infant - the infant selected whom to tell what to, often on the basis of the other's previous Tknowledge'. It is argued that these discriminations do not involve some formal conception of knowledge; rather they reveal a developing understanding of the enactment of knowledge in everyday life.

In other words, other people's knowledge is a part of their actions, and is a part of the infant's field of interactions. It is through engagement with people that infants come to act upon people's knowledge. And it is through such actions that formal conceptions may later emerge. But formal conceptions are not necessary for knowing about knowing.

Characterising Reactive Systems
Chris R. Roast
School of Computing and Management Science, Sheffield Hallam University

This paper presents a system model of user action which is to enable the expression of user oriented requirements in terms of system constraints. The approach benefits system development by ensuring that user oriented requirements are introduced, focused upon and formalised early within conventional software development. By expressing requirements in terms of a system model we are able to characterise a notion of action as supported by the system. This research builds upon previous work on the use of general system models within human computer interface design [1, 2, 3].

The development of successful interactive systems is dependent upon understanding both the intended users and the intended environment of the system. Conventionally, such an understanding is obtained through modelling users' interactions with a particular system, to complete well defined tasks within their working environment. The benefit of such analysis is tempered by its often specific nature: it fails to accommodate the variety of tasks a system may be used for, and; it does not generalise over the particulars of a chosen computer system. Thus, we suggest that the variety of users, tasks and working environments that computer systems 'encounter' cannot be realistically modelled.

As an alternative to understanding the circumstances of an interface's use, we propose to develop an understanding of behaviours appropriale for a variety of interfaces, tasks and working environments. This approach enables us to focus upon the computer system capabilities which support interaction in general. The basis of our approach is to view human computer interaction as an activity in which users act in an artificial environment that is defined by a computer system. Interaction involves users performing actions within the artificial environment, which, in turn, re-acts.

Successful interaction is based upon: (i) the system reacting in an informative manner, and (ii) enabling actions that assist with the users' manipulations. A system model of interaction is developed based upon this perspective, termed the template model ([3]). The template model introduces system abstractions, termed templates, that represent objects within the application domain and users perceptions of system output.

At the simplest level, the model identifies a set of system inputs (I) where any history of such inputs identifies a system state reached bv those inputs. Hence, we write statep and viewp to denote system state and system output after the input sequence p I*, respectively.

We can illustrate the use of this model by considering a simple requirement appropriate to any interactive system: An interactive system should respond to all possible inputs. This can be expressed in terms of the model as requiring that, for any input sequence p I* and single input i I, we require: viewp = viewp~(i).

The paper examines extensions to such simple forms of system constraint. In particular, we focus upon accommodating the variable perceptions of input, state and output which users may entertain. Using this we define a system view of action as, a sequence of inputs which result in an output response that is indicative of a change to the system state. A stronger form of action is also formulated, in which inputs to the system are treated as product of physical inputs (I) and display properties. In summary, this paper presents a formal system account of action based interaction. Such an approach is applicable to conventional system development methodologies and helps identify the extent to which systems can effectively support action based interaction.

[l] A. J. Dix. Formal Methods for Interactive Systems. Academic Press, 1991.
[2] M. D. Harrison, C. R. Roast, and P. C. Wright. Complementary methods for the iterative design of interactive systems. In G. Salvendy and M.J. Smith, editors, Designing and Using Human-Computer Interfaces and knowledge Based Systems, pages 651-658. Elsevier Scientific, 1989.
[3] C. R. Roast. Executing Models in Human Computer Interaction. PhD thesis, 1993. Department of Computer Science, University of York.

Coordinating Computer-Supported Work Activities: a Distributed Cognition Approach
Yvonne Rogers
School of Cognitive and Computing Sciences, University of Sussex.

Coordination of work activities is central to collaboration, especially when mediated by computer technology. However little is known about the practice of coordination in work activities: how it emerges, how it is maintained and what happens when it breaks down. The few empirical studies that have been carried have shown how complex and fragile it can be. Moreover, attempts at facilitating the coordination of joint work activities in organizations, through introducing new collaborative technologies, have tended to fail. The problem, it seems, stems from an insensitivity to what is going on in the real world.

If we are to develop computer-based systems that can enhance rather than hinder collaboration, then it is important to understand how existing work activities are coordinated and also how organizations attempt to adapt to new ways of working when new technologies are introduced into the workplace. In particular, a question that needs to be addressed is what happens when individuals who work together are required to coordinate their work even more in order to use new computer systems intended to support collaboration? Furthermore, how is the additional workload distributed amongst the people in the organization and what are the consequences?

In my talk I should like to examine these issues. Using distributed cognition as a theoretical framework, a recent ethnographic study will be presented, of an organization that is undergoing considerable change. The focus is on the coordination of the various actions and interactions of the people working in the organization. The implications of the analysis will be discussed in relation to the concept of distributed activity.

Body-scaledness and control in early prehension intention
Julie C. Rutkowska
School of Cognitive & Computing Sciences, University of Sussex

This paper draws on theoretical work about infancy, and on an empirical longitudinal study of prehension in 12- to 22-week-old infants. It is concerned with three broad issues about early abilities and their development: (1) How can we best conceptualize infant mechanisms? Contra recent emphasis on "naive theories" and networks of "(proto)concepts", I suggest that early object understanding is grounded in action, and that computational concepts -- of the right kind -- offer a useful route to discussing action systems and their openness to development; (2) What specific kind of organization underlies pre- adapted action structures? The example of prehension is used to discuss: the role of vision, in particular whether even inexperienced infants perceive objects in a body-scaled way, i.e. see graspability; and how motor processes may be organized and coordinated with vision, in particular whether infants attempt serial- ordering of pre-adapted motor components under the guidance of a controlling intention/explicit behavioural goal; (3) What kind of developmental mechanism can account for the way infants demonstrate immense regularities in outcome behaviour, yet remain capable of exploiting viable patterns of activity to establish novel coordinations? Even in the case of simple prehension, it is possible to see the routine emergence of activities that fall outside the restricted reach- graspretrieve-manipulate sequence against which most behaviour is interpreted, e.g. any manual contact with an object as "preparatory to grasping".

Much research on early prehension suggests it is supported by universal object properties such as surface-boundary segmentation, which vision could deliver through construction of a description in a representation akin to the 2\(12-D sketch of Marr's model of low-level vision. Specific properties such as shape or weight appear to be taken into account prior to contact only towards the second half of the first year. An exception is Bruner and Koslowski's (1972) classic finding that, prior to feedback from successful reaching, manual behaviour exhibits size discrimination between a potentially graspable object and a larger, non-graspable one. But problems with this work include its confounding of smaller/larger object sizes with the possibility of subjects seeing/not seeing objects' surface boundaries. A new longitudinal study questions interpretation of proximal midline activity of hands at the mouth as a "pre- reaching" manipulatory activity involving body-scaled appreciation of object size; and interpretation of anticipatory mouth opening as evidence for a "controlling intention" and attempted hierarchical control of pre-adapted component behaviours in a reach-grasp-retrieve-manipulate/mouth sequence. Current computational work that explores emergent functionality of independent sensory-motor coordinations (e.g. due to Brooks) is outlined, and its claim to be "noncomputational" is challenged. It is suggested that this approach to action organization provides a way of encompassing the notion that "intention precedes execution" without assuming explicit goals or hierarchical control. And it can be extended to a model of development that is pre-adapted without being predetermined.

Agreement and consensus
Wes Sharrock
Department of Sociology, University of Manchester

This discussion is a part of a systematic re- examination of the course of sociology of knowledge and science since its resuscitation under the influence of the so-called 'strong programme', a re-examination which reflects fundamental disagreement with 'constructionist' conceptions. One of the most influential sources of the 'constructionist' conception in the sociology of knowledge and science was Thomas Kuhn's The Structure of Scientific Revolutions and it is appropriate that such a re- examination should begin with a close inspection of Kuhn's views. This paper will consider the extent to which Kuhn's arguments are a product of his empiricist assumptions.

Action As It Appears in Russian Activity Theories and Ecological Approach
Alexander N. Sokolov
Institute of Psychology, Russian Academy of Sciences, Moscow, Russia.

The concept of action is one of the basic ideas in both traditional Russian psychology and the ecological approach. There are a few relatively distinct "action-oriented" approaches in Russian psychology (kinds of activity theory by A Leontiev, S Rubinstein, P Galperin, A Zaporozhets et al.) and physiology (physiology of activity by N Bernstein and theory of functional systems by P Anokhin). Similarly, one can discover several noncorresponding action-based theories in ecological paradigm, though for some time action per se seems to be outside the "key-words" of this approach. It is argued that the current general trend in both these theoretical frameworks consists in the shift from the use of action as a primarily explanatory tool to the consideration of action rather as a matter of empirical study (cf. issues of control of action, intersubjectivity). Some principal common features of action understanding by activity theorists and ecological researchers (eg, intentionality, orienting and adaptive function, importance of motor components/adjustment etc) as well as essential differences that are specific for both views (eg, dichotomy of external/internal actions in activity theory) are briefly outlined. Clearly, thorough analysis of these approaches' foundations is needed for they could enrich each other.

Metaphorical Activity in Communicative Context - a Dialogic Analysis of Figurative Language
Zazie Todd
Department of Psychology, University of Nottingham

This paper starts from the Bakhtinian premise that language is a social event in which both listeners and speakers play a very active role. Words and phrases gain their meaning from the dialogic situation in which they are uttered. With this in mind, figurative language is seen as a normal, creative use of the elastic environment of words we have at our disposal. This paper looks in particular at two questions, one theoretical, the other more directly related to language use. Does the concept of language - or communication - as a social event conflict with the idea, often expressed by ecological psychologists, that language is an index? Is it possible, in fact, for language to be a social index? And secondly, why do we use different types of figurative language? Does, for example, metaphor say something more - or different - to simile? How does the context, both social and situational, affect our choice of words? Preliminary evidence is presented concerning pre-school children's preferences for metaphor and simile, but this question relates to metonymy, synecdoche etc. as well. An examination of the social functions of figurative language may give us some insight into the transformation of meaning that seems to occur with metaphor. Indeed, it could be argued that this is just an extreme example of the negotiation of meaning that occurs every time a word or phrase is used.

ISEP HOME PAGE