On the Role of Social Interaction in Individual

|
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
 6 views
of 17

Please download to get full document.

View again

Description
On the Role of Social Interaction in Individual Agency 1 Hanne De Jaegher1,2, Tom Froese2 University of Heidelberg, Department of General Psychiatry 2 University of Sussex, Centre for Computational Neuroscience and Robotics Is an individual agent constitutive of or constituted by its social interactions? This question is typically not asked in the cognitive sciences, so strong is the consensus that only individual agents have constitutive efficacy. In this article we challenge this methodologi
Share
Tags
Transcript
  1 On the Role of Social Interaction in IndividualAgency Hanne De Jaegher 1,2 , Tom Froese 2 1 University of Heidelberg, Department of General Psychiatry 2 University of Sussex, Centre for Computational Neuroscience and Robotics Is an individual agent constitutive of  or constituted by  its social interactions? This question is typicallynot asked in the cognitive sciences, so strong is the consensus that only individual agents have con-stitutive efficacy. In this article we challenge this methodological solipsism and argue that interindivid-ual relations and social context do not simply arise from the behavior of individual agents, butthemselves enable and shape the individual agents on which they depend. For this, we define thenotion of autonomy as both a characteristic of individual agents and of social interaction processes.We then propose a number of ways in which interactional autonomy  can influence individuals. Thenwe discuss recent work in modeling on the one hand and psychological investigations on the otherthat support and illustrate this claim. Finally, we discuss some implications for research on social andindividual agency. Keywords agency · autonomy · cognitive gap · coordination · modeling · participatory sense-making ·perceptual crossing · social interaction · social cognition 1Introduction Is an individual agent constitutive of  or constituted by its social interactions?Our scientific understanding of individual agencyhas come a long way since the inception of the cogni-tive sciences in the early 1970s. What started out witha focus on computational architectures in the heydayof cognitivism has been supplemented with a connec-tionist concern for self-organization and emergence inthe 1980s, and grounded in an embodied, situated, anddynamical perspective of cognition in the 1990s. Thischange in our understanding, which coincides with ageneral shift toward privileging the concrete over theabstract (Varela, 1995), has produced some significantadvances. In particular, it has become more widelyacknowledged that there exists some sort of continuitybetween the phenomena of life and mind (e.g., Di Paolo,2003, 2009; Stewart, 1996; Thompson, 2007; Wheeler,1997). However, this life–mind continuity thesis hasnot yet culminated in a fully worked out theory of individual agency and cognition. Most theorists whosubscribe to some version of this thesis recognize that animportant challenge remains: how does such an accountcross the gap between insect-level behavior, which hasreceived a lot of attention since the early 1990s (seeBrooks, 1991), and human-level cognition, which hasbeen the traditional focus of mainstream cognitive sci-ence? We will refer to this particular problem of thelife-mind continuity thesis as the cognitive gap .Why has it been so difficult to conceive of howwe can apply the insights that have worked so well for Copyright © 2009 International Society for Adaptive Behavior(2009), Vol 00(0): 000–000.DOI: 10.1177/1059712309343822 Correspondence to : Hanne De Jaegher, Projekt DISCOS, Klinik fürAllgemeine Psychiatrie, Zentrum für Psychosoziale Medizin,Universitätsklinikum Heidelberg, Voß-Str. 4, D-69115 Heidelberg,Germany.  E-mail : h.de.jaegher@gmail.com. Tel. : ++44 6221 5637411, Fax : ++44 6221 568094  2 Adaptive Behavior 00(0)  models of minimal cognition (Beer, 1995, 2003) inorder to understand and engineer artificial forms of the human reaches of cognition? The most importantpoint to understand in this regard is that the cognitivesciences, even though they have undergone some radi-cal changes, are still largely dominated by methodolog-ical solipsism. That is, after the cognitive revolutionmade behaviorism a thing of the past, interest in therole of the environment was replaced by an almostexclusive focus on the internal capacities of an individ-ual cognitive agent. From this perspective the cognitivegap must indeed appear to be an insurmountable prob-lem: how can the simple mechanisms that realize insect-level behavior be essentially similar to the ones thatrealize the whole spectrum of possibilities that is char-acteristic of human agency and cognition? Is it just amatter of increasing the complexity of our models? Tobe sure, it has long been shown that behavior thatlooks intelligent to an external observer does not haveto arise from internal mechanisms that are equallyintelligent (e.g., Ashby, 1960). However, the differ-ence between insect and human cognition is so vastthat a simple life–mind continuity thesis appears to beimplausible (Brooks, 1997; Kirsh, 1991).Note, however, that this extreme version of the cog-nitive gap is based on what can be called an “internalist”approach to cognitive science, namely the idea thatagency and cognition are largely constituted by whatClark (2008) has aptly termed “brainbound” operations.To put it bluntly: if we accept the idea that our cogni-tive processes are essentially realized solely within ourbrains, then it indeed becomes impossible to conceivehow the explanatory framework that has worked sowell for insect-level behavior could do justice tohuman cognition. But what if we give up the commit-ment to an internalist view of the mind? This opens upthe possibility that external factors could also play aconstitutive role for higher-level cognition, and thuspave the way for a more moderate life–mind continu-ity thesis.The idea that the environment plays a constitutiverole for cognition has already been developed exten-sively in the cognitive sciences since the early 1990s.From robots that use “the world as its own best model”(e.g., Brooks, 1991), to dynamical accounts of howcognitive behavior emerges out of the dynamics of abrain-body-world systemic whole (e.g., Beer, 1995;Kelso, 1995; Thelen & Smith, 1994), to analyses of howenvironmental structures can provide “scaffolding” forcognitive problem solving (e.g., Clark, 1997, p. 45;Hutchins, 1995), and the hypothesis of extended cogni-tion and the “extended mind” (e.g., Clark, 2008; Clark & Chalmers, 1998), the role of the environment isclearly back on the agenda. The concepts of scaffoldingand the extended mind are of particular interest herebecause they deal with external factors that are specifi-cally constitutive of human cognition, especially tech-nology and language. Indeed, by arguing for the claimthat human cognition is essentially a distributed phe-nomenon, they support a life–mind continuity thesisthat is not caught in the internalist trap. However, thisconnection between a potential reduction of the cogni-tive gap and the externalism promoted by these embod-ied–embedded theorists has not been made explicit intheir writings. The focus on technology and language,as well as human culture more generally, has obscuredthe fact that the principles at work in these cases canalready be found in much more basic forms of agency,cognition and interindividual interaction.In this article, accordingly, we will use the enac-tive approach to social cognition in order to highlightways in which individual cognition and interindivid-ual interactions mutually enable and constrain eachother. From this perspective it will appear that thesetwo aspects of human agency are intrinsically linkedand must both be taken into account by researchersinterested in modeling our cognitive capacities. Theinteraction examples discussed in this article do notrely on the specifically human capacities for languageand technology, but rather on lower-level phenomenasuch as coordination. For this reason, they lend them-selves to exploration by modelers interested in thisdirection of research.We begin, in Section 2, with a brief discussion of some central concepts and their definitions: autonomy,adaptivity, agency, and sense-making. Our purpose isnot to justify these definitions in their full detail—thishas been done in the work we refer to. Nor do we wantto say that roboticists and modelers should adherestrictly to them. One thing that will become immedi-ately clear is that these definitions are very demand-ing: So demanding, in fact, that they are not applicablein the current research context. Today’s robots are notautonomous in the sense of the word as we use it here.We do think, however, that practical implications canbe drawn from them. These definitions can help guideresearch in certain directions, even if work done ontheir basis would seem to overshoot the current mark.  De Jaegher & Froese  Social Interaction and Individual Agency 3 So while we think the concepts of autonomy, adaptiv-ity, social interaction, and sense-making as definedbelow are not directly applicable to existing instancesor models in robotics, we hope they can provide bothsome clarity as to the target that this research couldaim at, and a set of practical sub-questions that couldbe addressed by the techniques of adaptive behaviorresearch. It is the aim of this article to work out some of these implications and guidelines in the area of socialinteraction. Therefore, in Section 3 we review recentarguments for a natural extension of the enactive frame-work into the social domain in terms of participatorysense-making. In Section 4, we discuss how this opensup the notion of individual sense-making to interindi-vidual influences. In Section 5 we discuss examples of modeling and experimental work that back up andillustrate this theoretical backdrop, which allows us tofinally conclude that social agents, as well as beingconstitutive of the social interactions they engage in,are also constituted by them. 2Agency and Sense-Making Since the enactive approach to individual agency is thefocus of discussion for several other contributions of this special issue (e.g., see Barandiaran, Di Paolo, &Rohde, in press, this issue), here we will only examinedefinitions of the most important concepts. Later on,we use these to draw implications about the way inter-actional and individual elements constrain and influ-ence each other.One of the key concepts that ground the enactivework on autonomous agency is that of  autopoiesis , asystemic characterization of metabolic self-production(Varela, Maturana, & Uribe, 1974). However, sinceautopoiesis is usually reserved for autonomy in thechemical domain we need a different concept foroccurrences of autonomy in other domains. Accord-ingly, Varela put forward the notion of  organizational(or operational) closure by taking “the lessons offeredby the autonomy of living systems and convertingthem into an operational characterization of  autonomyin general , living or otherwise” (Varela, 1979, p. 55).He says thatautonomous systems are organizationally closed.That is, their organization is characterized by proc-esses such that1.the processes are related as a network, so that theyrecursively depend on each other in the genera-tion and realization of the processes themselves,and2.they constitute the system as a unity recognizablein the space (domain) in which the processesexist. (Varela, 1979, p. 55)In order to separate this definition of the term auton-omy from other typical uses found in artificial life andthe cognitive sciences, we specifically distinguish it as constitutive autonomy (see Froese, Virgo, & Izquierdo,2007). This concept is fundamental to the enactiveapproach for several reasons. First, it enables us to talk about processes of  identity generation (see Di Paolo,2003), that is, by being organized so as to continuallycreate itself, the operation of the system also defines theway in which it is organized. Second, the mutualitybetween constitutive factors in this identity generationmakes it possible for us to attribute intrinsic teleology tothe system itself (see Weber & Varela, 2002). Thismeans that the relations of cause and effect are at thesame time relations of means and purpose. Third,because this identity is an accomplishment of the sys-tem, rather than something pre-given or imposed fromthe outside, it naturally finds itself in a  precarious situa-tion that requires continuous overcoming (see Di Paolo,2009). In order to effectively deal with the precari-ous situation of their identity generation, living sys-tems need to be able to behave in an adaptivemanner. This additional property of  adaptivity can bedefined as follows:A system’s capacity, in some circumstances, toregulate its states and its relation to the environmentwith the result that, if the states are sufficiently closeto the boundary of viability,1.tendencies are distinguished and acted upon depend-ing on whether the states will approach or recedefrom the boundary and, as a consequence,2.tendencies of the first kind are moved closer to ortransformed into tendencies of the second and sofuture states are prevented from reaching theboundary with an outward velocity. (Di Paolo,2005, p. 438)Thus while self-constitution establishes a purpose anda perspective, adaptivity is needed so that the system  4 Adaptive Behavior 00(0)  can anticipate and act upon that which it needs for itsself-constitution and that which may threaten it.Systems that are both autopoietic (or, more gener-ally, constitutively autonomous) and adaptive arecapable of  sense-making (see Di Paolo, 2005; Weber& Varela, 2002). Sense-making is the capacity for asystem to enact a world of meaning. 1 This entails thatthe system is able to interact with the environment interms of the consequences that its interactions havefor the conservation of its identity. In other words,such consequences have significance or value for thesystem, namely in relation to the processes of its iden-tity generation. This is because self-generation is asource of norms: different events will contribute dif-ferently to its continuation, some enhancing it, othersputting it in danger.This self-given normativity applies at the meta-bolic level and all the way up to the highest kinds of identities that human beings can make for themselves.At the direct metabolic level this means that certainelements of the environment count as nutrients andothers do not. But it can also mean that through adap-tivity the system is able to make sense of more indi-rect forms of coupling with the world and sustain therequirements of several organic norms. Such is the casefor migrating species that travel long distances to findnourishment when, for instance, breeding grounds arefar away from the main source of food, enduring longperiods of hunger and navigating through difficultlandscapes in the mean time.In this view, the idea of sense-making is inti-mately connected to the conservation of a precariousidentity. Metabolism is the most basic example, butin a given organism it is possible for several suchprocesses to overlap, bringing forth the possibility of non-metabolic norms such as habits of behavior orsustained forms of relations within a group. In thehuman ranges, norms of interaction with the worldcan go so far as to put severe constraints on metabolicidentity, for instance in extreme dieting. A socialidentity also needs to be preserved. When a personrisks getting fired from her job, her identity as anemployee is in danger of breaking down, but thisdoes not necessarily have an immediate effect on hermetabolism.An important aspect of sense-making is the cen-tral role of  movement  in it. As Sheets-Johnstone (1999)says, agents make sense of the world in movement.Even though Sheets-Johnstone’s approach is primarilyphenomenological and in the first place applicable tohumans, this principle is well-known in adaptivebehavior research. For instance, Beer (2003) describesan evolved agent capable of discriminating objects interms of their shape through active scanning, asopposed to some kind of internal reconstruction of thesensory pattern followed by appropriate calculationsto generate the categorization. Thus, the process of artificial evolution found a solution to the task thatuses a pattern of moving exploration. While this agentis not autonomous, its behavior illustrates the role of movement in aspects of sense-making (for the role of moving exploration in cognition, see also O’Regan &Noë, 2001).In short, autonomous adaptive systems enact aworld of meaning and value through their movementin it. Sense-making, then, is the capacity of a systemto enact a world and imbue it with significance fromits own point of view. An autonomous adaptive sys-tem does this in relation to the processes of identitygeneration that realize it as the particular organism-environment systemic whole that it is. We can nowdefine an agent  as any autonomous system capable of sense-making in its interactive domain (see Barandi-aran et al., in press).This definition of agency does not tally with theway the notion is used in most of robotics researchtoday (see Froese & Ziemke, 2009). According to thepicture of autonomy, adaptivity and sense-makingsketched here, no robot with these “qualities” exists asyet. Robot behavior can conform to an externallyimposed norm, but robots still do not produce normsthemselves (Di Paolo, 2003) and this is currently themain obstacle for calling these robots agents in thesense of the word espoused above. Nevertheless,robotic modeling can contribute to our understandingof these concepts by studying specific sub-aspects of these phenomena, without necessarily having to rep-licate them in their entirety. For instance, it is possi-ble to model phenomena such as social interactioneven if agency in the sense above cannot yet bemodeled, as we will see in Section 5. 2 But beforeshowing practical examples, we discuss a theoreti-cal extension of the enactive ideas into the socialdomain in order to get a first grasp of how agencyand sense-making might be affected by interindivid-ual interactions.
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks