The Assumption of Agency Theory

Kate Forbes-Pitt

• Publisher: Routledge
• Year: 2011
• Print ISBN: 9780415782111
• eBook ISBN: 9781136633256
• Topic: Social Sciences
• Subtopic: Sociology
• Index: Social Sciences

The Assumption of Agency Theory revisits the Turing Test and examines what Turing's assessor knew. It asks important questions about how machines vis à vis humans have been characterized since Turing, and seeks to reverse the trend of looking closely at the machine by asking what humans know in interaction and how they know it.

Building upon existing theories of philosophy of mind, this book shows not how humans operate theoretically, but how they use every day human skill to overcome knowledge barriers and understand each other through knowing themselves. Only once human interaction has been theorized in this way are machines able to be placed within it; when it is easier to understand what humans believe them to be. This book characterizes a non-human agent that shows itself in interaction but is distinct from human agency: an agent acting with us in our ongoing reproduction and transformation of structure. Turing predicted that at the end of the twentieth century, we would refer to thinking machines 'without fear of contradiction'. The Assumption of Agency Theory shows how and why, even if we don't say it, we deal with machines every day as if they are thinking, acting agents.

1 Philosophical Assumptions

The purpose of this chapter is two-fold: first, to lay out in a clear and unambiguous manner the philosophical assumptions that underlie this book. Doing so at this stage does not mean that the issues raised throughout what follows that directly reference, or perhaps challenge, these philosophical assumptions will not be discussed at the relevant points. Neither does it mean that they will not be revisited where appropriate. One aim of this initial outline and discussion of both the assumptions at play throughout what follows, and the reasons for rejecting others, is to reduce doubt where the assumptions are not made explicit in the remaining text. This also obviates the need, in later chapters, to break the flow of the argument to introduce ideas in their philosophical context.

Second, it establishes a sound metatheory on which to base the theoretical propositions that follow. To be mistaken, unsure or unclear about these assumptions would be damaging to the strength of the following arguments. My aim in this book is to gain a deeper understanding of human beings, their interaction and how agency is produced within this. Issues such as the status of ‘fact’, the primacy or otherwise of language, the necessity or otherwise of shared meaning, the direction of causality and the distinction of subject and object within the philosophical assumptions that underpin the theory might initially appear arcane. However, these are not small or trivial details but fundamental to good theory and sound theoretical propositions: these assumptions form the bedrock of theory building. I will argue later, for example, that humans and objects are separate; independent of each other. Establishing a foundation, then, in which the subject and the object are separate and the dominance of entity over language is established, and not the ontological exaggeration that denies this (Fleetwood, 2005), is not indulgent but necessary. I will also discuss at length the concept of ‘mind’ and the possibility of first- and third-person knowledge of it. Thus, it is sensible at this stage to at least mention a view that denies meaning to first-person knowledge of mind and to be explicit about the disagreement.

I will also make an argument for a new category of ontologically real later in the text. It is incumbent upon me, then, to establish that there are existing categories of ontologically real, that these are not conflated with representations of objects and that the ontic nature of such objects is unchanged by their relationship to discourse. What follows in this chapter aims to both establish some clarity in these matters for the reader as they are taken forward and describe the basic philosophical foundations for my following arguments.

Ontology describes what is; the nature of something. The basic ontological assumption throughout this book is realist and so the basic underlying supposition is that there is a reality external to those perceiving it in which objects exist independently of intervention by them. Epistemology is the theory of knowledge; knowing about what is. The basic epistemological assumptions contained here are based in human perception.

The specific way in which these two positions are presumed to be related is in the fundamental assumptions of critical realism. Critical realism is not the only philosophy to bring together realist ontology and epistemology based in perception. Campbellian realism (McKelvey, 1999), which utilises evolutionary epistemology, has been proposed as another way to unite these two basic approaches. Within this book, however, I do not assume evolutionary epistemology, in fact I take issue with it at points; the assumptions here are based specifically upon Bhaskar’s transcendental realism.¹ It also utilises Fleetwood’s (2005) helpful four-fold ontology to elucidate the ontological status of the real entities it proposes.

Critical realism is distinct from those versions of postmodernism and poststructuralism that subscribe to a strong social constructivist ontology. In what follows, ‘postmodern’ is used to describe both of these positions in a number of ways, but principally with regard to ontology. Within strongly social constructivist versions of postmodernism, the ontological status of an entity is decided by its involvement, or otherwise, in discourse. That is, entities are constructed, created, produced or whatever via discourse, so that only those entities constructed in this manner are real. More realist looking versions of postmodern ontology attempt to hold on to two ontological realms: the extra-discursive realm and the discursive realm (Laclau and Mouffe, 1987; Alvesson and Deetz, 1999). An entity can exist in either realm, depending on its relationship to discourse. States are associated with these realms; ‘existent’ and ‘being’ respectively. All entities exist in the extra-discursive realm as ‘existents’ until they become the subject of discourse. At this point the entity both moves into the realm of discourse and changes state from ‘existent’ to ‘being’. This does not create ‘parallel’ entities, but the same entity is moved between realms. So, even to identify an entity is to change it from ‘existent’ in the extra-discursive realm to ‘being’ in the discursive realm. It follows that nothing can be said about existents in the extra-discursive realm without transforming them: they are changed into ‘being’ by becoming a subject of discourse.² (See Fleetwood, 2005 for an elaboration of these versions of postmodernism.)

The notion of an entity’s ontological status being changed as a result of discourse is rejected throughout what follows. This will become particularly important in Chapter 3 when discussing Searle’s intentionality and mental representation.

Critical realism utilises realist ontology. Unlike postmodern ontology, critical realism recognises no change to the ontological status of an object as a result of discourse – unless the discussion is about discursive entities themselves, which I will shortly define as ideally real. Within critical realism, an object or entity is said to exist independently of any attempt to bring it into discourse, even after it has been brought into discourse:

Saying an entity can exist independently of its identification implies that is can exist without someone observing, knowing and constructing it.

(Fleetwood, 2005: 199)

In other words, within critical realism discourse does not act as a distinguishing factor in the nature of an entity. The reality of an object or entity is decided on the basis of whether or not it is causally efficacious: in other words, whether it makes a difference. This includes, though is in no way limited to, material entities. It is important to note that critical realism does not assume any alignment of ‘real’ with ‘material’ or ‘non-material’ with ‘non-real’.

In critical realism, all entities are ontologically described using what Fleetwood calls ‘modes of reality’. In addition, critical realism uses conceptual mediation to allow for human perception, concurring with one implication of postmodernism; there is no possibility of either direct or theory-neutral observation of the world. In this respect, Bhaskar (2008 [1975]) upholds the Kantian concept of the veil of perception within critical realism; an acknowledgement of the role of perception as a mediating factor in humans’ interactions with the entities in the world making direct observation impossible qua empiricism. The notion of conceptual mediation, then, allows naive and empirical realism to be avoided by not conflating experience with reality, and allows the richness of human perception of entities to be introduced into realist ontology: ‘When they become the focus of human beings’ reflection, then, entities may be said to be conceptually mediated’ (Fleetwood, 2005: 199).

In critical realism, an entity can be materially real, ideally real, artefactually real or socially real (ibid.). Materially real entities are things that exist wholly independently of human existence or human experience of them: oceans, mountains and the planets, for example. This is not to say that there cannot be causal effect either way between humans and materially real entities. What it does say is that these entities would exist without humans and do not require human identification or interaction in order to exist. These entities, however, are often identified, and when they are, they are conceptually mediated. It is important to note that the conceptual mediation does not change the ontological status of the materially real entity in any way. It becomes a conceptually mediated materially real entity: an entity with a co-existing identification.

Ideally real entities are conceptual – e.g. belief, meaning, ideas – but they are not conflated with conceptual mediation. Ideally real entities cannot be observed, but they have no less causal efficacy than materially real entities.

Ideally real entities can have referents, or they can have no referent. Discourses about mountains have as their referent the materially real entity ‘mountain’. Discourses about an ideally real entity – such as discourse about postmodernism discourse – have the ideally real entity as their referent. Just as with materially real entities, the ontological status of the ideally real entity itself is unchanged by its identification – although such an entity ‘just is’, as it were, constituted by ideal stuff. Note that we ought not to confuse identification and interpretation, which is a sub-set of identification. An entity is not changed by interpretation, hence allowing the possibility of mis-interpretation.

Socially real entities are things like self-organising social structures:

Critical Realists use the term ‘social structures’ as a portmanteau term to refer to configurations of causal mechanisms, rules, resources, relations, powers, positions and practices.

(Fleetwood, 2005: 201)

Socially real entities have no materiality and depend on human activity for existence. This dependency might suggest that they have no existence outside human identification, but this is not the case; socially real entities can pre-exist their discovery. Class and gender relations, for example, existed before we identified them as such. In fact, almost every kind of prejudice can be argued to have had causal efficacy – and so to have existed – prior to its identification. Again, the involvement in discourse of socially real entities will not change their reality; it will simply make them conceptually mediated. Thus, the socially real entity can be seen to pre-exist identification and be independent of it. Importantly, no explanation, identification or theorisation of the socially real entity changes its ontological status – although there is one important caveat. It is possible that, say, discourse about a socially real entity like gender relations can be changed by discourse, perhaps non-sexist, about it. Much depends upon the particular entity and the context and the efficacy of the discourse, but this possibility cannot be ruled out. Perhaps such discourse might bring into existence a co-existing ideally real entity, or simply make the socially real entity conceptually mediated. Sometimes other modes of reality synthesise in objects to form artefactually real entities; computers, for example, synthesise the materially, ideally and socially real and are artefactually real.

Entities can be perceived in diverse ways and critical realism does not preclude limits on interpretations:

Violins may be interpreted as musical instruments or table tennis bats. But, unless we are prepared to accept that any interpretation (and therefore subsequent action) is as good as another, that interpreting a violin as a table tennis bat is as good as interpreting it as a musical instrument, then we have to accept that there are limits to interpretation.

(Fleetwood, 2005: 201)

To illustrate ideally real entities, materially real entities and existence pre- and post-human intervention, consider the astronomical entity Pluto. Pluto is a materially real entity, it requires no human intervention to exist and will very likely carry on existing long after humans have ceased to exist and call it Pluto (and what is there exists, however humans choose to classify it). Pluto wasn’t discovered until 1930, though clearly it existed before this date. Prior to its discovery, there were theories about the existence of an additional planet as it was thought to be the only way to explain anomalies in the orbits of Uranus and Neptune. In fact, the predictions of another planet’s existence were flawed, but this did not come to light until the Viking spacecraft discovered Neptune’s true mass in 1984. However, on the basis of the flawed theoretical prediction, Clyde Tombaugh discovered Pluto in 1930.

Pluto, then, is a materially real entity that existed prior to its discovery and its identification by humans. The theory that predicted the existence of another planet had causal efficacy; it caused the materially real entity to be discovered and identified. Thus, using critical realist ontology, the theory of the existence of another planet can be shown to be an ideally real entity: it was conceptual and had causal efficacy.

The verisimilitude of the ideally real theory could be ascertained by measuring the causal mechanisms of the materially real entities it referenced. Because of this measurement, the various ideally real entities referencing the materially real entities were subject to justification logic, and so verisimilitude (or otherwise) could be established. This is illustrated by the theories that were in use in the discovery of the planet being shown to be ‘nearer’ or ‘further’ from what is seen as correct information about the entities upon which they were founded; the mass of Neptune, for example.

This example shows the strength of critical realism. It accounts for the reality of the causally efficacious entities under discussion and it allows for their perception and multiple interpretation through conceptual mediation, without altering their ontological status. Because the reality of the entities is not denied – or altered by discourse – they are potentially subject to measurement (at least those that lend themselves to this operation) and so justification logic; verisimilitude can be established. As opposed to the effect of an ontology that does not allow for independent existence and in which measurement is all relative to a given relativisor. Where this is the case, no measurement independent of the relativisor can be seen to exist. Thus, there could be no ‘true’ mass for Neptune, for example, and so the flawed nature of the theory could never be established. In this way, the possibility of justification logic is denied, as is the possibility of verisimilitude, and nothing can actually be said about the entity itself, merely the relativisor’s perception of it. Whereas in critical realism the perception is relevant but potentially subject to correction, in postmodernism it is not possible to see anything but the perception. This does not mean or imply that verisimilitude is always possible with the critical realist framework – there are complex matters to address such as questions involving the technical efficacy of measuring instruments and the relationship between ideally real and materially real entities. However, it does imply that verisimilitude is not possible where the ontic nature of something is uncertain and can only be measured relative to a relativisor. Such a position has fundamental consequences for causality and I hold that the possibility of verisimilitude, if not its attainment, is important.

The notion of causality comes under discussion in what follows, and again a critical realist understanding of what underlies this concept is assumed: that entity precedes language. I believe this is crucial and that it is important to be explicit about this relationship at this stage. The causal relationship between language and entity will become important in Chapters 3 and 4.

Fleetwood (2005) uses du Gay’s (1996) own example of a stone to disambiguate the nature of ‘fact’, within postmodern ontology and its subordination to language, or otherwise. This is a useful way to begin a discussion of the effects of differing ontological assumptions on the nature of causality.

Fleetwood (2005) is paraphrased as follows: ‘fact’ might be taken to refer to an entity or the classification of an entity: X is a stone is a fact (fact refers to entity) vs X is classified as a stone is a fact (fact refers to classification). In the former, if fact is subordinate to language, as postmodernism contends, then entity is subordinate to language and thus language must bring entities into being [i.e. language is causally prime]. In the latter, if fact is subordinate to language, then facts are linguistic entities and only these are subordinate to language, not the entities themselves [i.e. entities are causally prime and language has causality only over linguistic entities]. Three possibilities arise from this argument:

1. Language is subordinate to entities: entities are a precondition of language (the traditional approach).
2. Entities are subordinate to language: language is the precondition of entities derived from the first approach above (X is a stone is a fact).
3. Facts are subordinate to language: language is the precondition of facts derived from the second approach above (X is classified as a stone is a fact).

Postmodernists reject (1) ‘it gets the direction of causality between language and entities the wrong way around’ (ibid.: 211). To replace (1) with (2) solves the problem, but the result is to say that entities are brought into being by language [an illustration of the primal causal role given to language by this argument]. Thus (3) is...

Table of contents

• Cover Page
• Halftitle Page
• Series Page
• Title Page
• Copyright Page
• Table of Contents
• Preface
• Acknowledgements
• Introduction
• 1 Philosophical assumptions
• PART I Self-knowledge and the assumption of agency
• PART II Interaction, maintenance of agency and the unexplained act
• PART III Technology as agent
• Notes
• Bibliography
• Index