Structures,Agents and Criticism: Assessing Bhaskar's Fact-To-Value and Value-To-Fact Arguments

Introduction

Roy Bhaskar's (1991, 1997 and 1998) critical realism is becoming an increasingly influential approach in political science. Bhaskar argues for a non-positivist science of politics. Positivism is held to commit the ‘epistemic fallacy’ (Bhaskar, 1997, p. 16) of reducing questions about reality into questions about how we can know reality: what exists becomes what can be observed through measurement. The consequence of this is that correlations between variables in quantitative research are taken to be causal relations. The problem with this is that correlations cannot be treated as causal relations because correlations are contingent and break down (Bhaskar, 1998; Sayer, 1992).

The realist alternative is to shift attention to structures that are unobservable in themselves but that condition (rather than determine) the behaviour of agents (Bhaskar, 1998). These structures are held to be emergent properties because while they stem from the actions of individuals, they have causal power over individuals. Thus capitalism, for example, stemmed from the actions of individual traders but is now a structure with causal powers that are irreducible down to individuals. This is not to imply that structures drive change, for agents may change structures, or one group of agents may preclude another group from changing a structure, or structures may resist change from agents. The task of political science thus becomes that of explaining observed events in terms of the interaction between agents and structures. All of which means that structure and agency becomes an important problem to be addressed in political science, along with the problems of linking theory to methods, explaining how and why conflict sometimes produces change while at other times it results in continuity, and assessing whether or not political science should be critical or value neutral.

In contrast to positivism, which holds that objective research must be value neutral, Bhaskar advocates the need for criticism. Bhaskar gives us two arguments about criticism. First, the fact-to-value argument for ‘explanatory critiques’ holds that we must logically derive a negative value judgement from a fact about structures that cause false beliefs. This argument is taken to show that the naturalistic fallacy is a non-fallacy, by refuting those from Hume (2000) on who argue that values cannot be logically derived from facts. Second, the value-to-fact argument holds that any ‘descriptively adequate’ account of facts that are ‘value impregnated’ with the values of agents must use normative language, otherwise it will distort the facts and not be objective. So, criticism can be objective because it turns on logically deriving value judgements from facts about structures and describing some facts accurately using value impregnated language to reflect agents’ values. This is meant to give us a realism that is also critical.

Most of the literature on critical realism deals with developing or criticising the notions of structure and agency and their application to empirical research: see for instance the exchange between Stuart McAnulla (2005) and Colin Hay (2005) in this journal. The argument developed in this article, though, contributes to the growing literature on critical realism by addressing Bhaskar's argument for objective criticism. It is argued that any critical non-positivist political science needs to be based on the value-to-value argument, whereby value judgements stem from prior rationally held value commitments and not a fact-to-value or value-to-fact argument.

From facts to values

The fact-to-value argument concerning the logical derivation of values from facts is based on ‘explanatory critiques’ of ideology, with ideology defined as beliefs that are both false and caused by a structure. Explanatory critiques explain how a structure causes false beliefs and are critical for three reasons. First, it is held that the very act of criticising a belief system for being false will also be an act of criticising both the actions informed by those beliefs and any structure that causes those beliefs (Bhaskar, 1998, p. 63; see also Bhaskar, 1991, p. 151). Second, negative value judgements may be made about the structure that causes false beliefs. Third, positive value judgements may be made about any rational action, ceteris paribus, which is geared up to remove the structure that causes false beliefs (Bhaskar, 1998, p. 63; see also Bhaskar, 1991, p. 151).

The making of such value judgements is an act of logical derivation because, for Bhaskar, both factual statements and normative statements about values are premised on a commitment to truth. Anticipating the criticism that his starting point is actually a value judgement that one ought to value truth over falsity, Bhaskar argues:

With this view the nature of the commitment to truth would have to be epistemic rather than normative because the starting point would be with truth and not a normative commitment to value truth. One would be committed to truth because truth was self-justifying and not in need of any normative commitment: a true statement is accepted as such because it is true. Thus in normal conversation one does not ask for a further justification of a statement if it is accepted as true: one does not, ceteris paribus, say ‘I accept your argument that X is true but why should I accept it?’

However, if we agreed that the starting point was with an epistemic commitment to truth and not a normative commitment to value truth over falsity then this case could not be stated. For if one follows Bhaskar and states that a commitment to truth is the condition of all discourse then one is making a factual claim, namely that it is a fact that a commitment to truth is the said condition. The problem here is that the commitment to truth is meant to be antecedent to any factual claim. One cannot seek factual status for the claim that the commitment to truth is the condition of factual statements without hitting the problem that the commitment to truth comes before any factual claims, including the claim that it is a fact that truth is the condition of all meaningful discourse. So, Bhaskar gives us the paradox that accepting his argument entails rejecting it because one could not make such a factual claim about the commitment to truth. As the fact-to-value argument hits this logical problem we may argue that it is more tenable to base criticism of ideology on the value-to-value argument, whereby criticism stems from a normative commitment to value truth over falsity.

From values to facts

Bhaskar also argues that we can move from values to facts. Bhaskar assesses two types of value-to-fact argument. The first focuses on the social scientist and holds that knowledge claims may have their objectivity compromised or prevented altogether by the value bias of the social scientist. The second focuses on the fact or object of knowledge, with the argument being that the object is ‘value impregnated’ by agents’ values, and so any objective account that seeks ‘descriptive adequacy’ of the object of study must entail evaluative language.

Bhaskar criticises three arguments concerning the value bias of the social scientist. First, Bhaskar holds that those working in the applied sciences may have ‘practical’ (funding, career, ideological, etc.) interest in seeing a project produce the desired results but ‘pure’ science will work independently of such interests in getting knowledge of underlying structures (Bhaskar, 1998, pp. 55–56). Seeking knowledge of structures may be different from pursuing government-funded research into security policy outcomes, for example. So, it is not necessarily the case that values in the shape of practical interests will distort the findings of all the sciences. Second, Bhaskar considers the argument that objective research needs to be value neutral, and that to assist this the researcher needs to make clear any values or biases they may have. On this position, Bhaskar argues that one either achieves total self-awareness, which is unlikely and which makes the statement of any potential value bias redundant, or that one is unable to become conscious of all the possible unconscious and subconscious biases one has, in which case statements about one's values and biases are worthless (Bhaskar, 1998, pp. 56–57).

Third, Bhaskar considers the position that there is ‘a relativity in the methodological norms secreted by different conceptual schemes or paradigms, together with a value dependence of such conceptual schemes’ (Bhaskar, 1998, p. 57). Bhaskar responds to this by arguing that we need to distinguish judgemental relativism from epistemic relativism. Whereas the former would hold that there are as many truths as there are methodological protocols or conceptual schemes, the latter holds that there are different interpretations of reality, with some having more truth content than others. Bhaskar argues that epistemic relativism is entailed by realism because if we cannot directly observe the functioning of structures then truth claims are fallible and we must seek better interpretations of the world which have greater truth content than currently held interpretations (Bhaskar, 1998, pp. 57–58). As epistemic relativity is taken to be entailed by the realist approach to the sciences it is taken to be correct. Accepting epistemic relativity means accepting that our knowledge will be fallible but this is different from holding that our perception of reality is distorted and thus undermined. So, for Bhaskar, value bias may exist and distort the data, but it is only really a problem in the ‘applied’ sciences. The ‘pure’ political scientist seeking better (fallible) interpretations of structures is not deemed to be at risk of distorting the data with personal value biases.

Turning now to the argument about the value impregnation of some facts, Bhaskar holds that ‘the subject-matter of social science is itself in part constituted by, or indeed just consists in, values or things to which the agents themselves attach … value, that is, objects of value’ (Bhaskar, 1998, p. 58). As some facts – or objects of study – are ‘value impregnated’ those facts will have to be described using value-laden terminology to provide descriptively adequate accounts of the objects in question. If one tries to describe some facts in a value-neutral way one will not be giving an objective account because one will be describing those facts incorrectly. Therefore we should not conflate objectivity into value neutrality because a properly objective account of some facts will be an evaluative account as well (Bhaskar, 1998). In making this case, Bhaskar draws upon Isaiah Berlin's discussion of different descriptions of the Holocaust, arguing thus:

So, in addition to criticising structures for causing false beliefs, an objective political scientist should also be able to describe some facts in a critical and value-laden way.

If we accepted this argument then we could say that the Holocaust is a fact that is impregnated with values by many agents who regard it as a massacre of millions and so an adequate description of the fact must use evaluative language. The problem though is that this ties descriptive adequacy to the meanings built into events, structures and institutions, etc. by agents. This may be acceptable if we agreed with agents' values but it would become a problem if the task was to criticise something of which many agents did not have a critical view. So, for example, if we wanted to use evaluative language to criticise xenophobia in a context where xenophobia was becoming increasingly popular, there would be a problem with describing this object of study in a way that was different from the values impregnated into it by agents. One could counter that in such an instance the commitment to truth ought to take priority over such descriptive adequacy, with an explanatory critique exposing such xenophobic beliefs as ideological. However, this takes us back to the problem that explanatory critiques could not logically derive values from facts. However, we could say that a normative commitment to value truth, anti-racism and anti-chauvinism motivated a criticism of xenophobic beliefs in agents. We could motivate the break from descriptive adequacy based on our normative commitments leading to critical normative judgements about such agential beliefs, combined with a description that differed from agential beliefs. This would, of course, give us a value-to-value argument.

Andrew Collier (1998, p. 449), by contrast, argues that the starting point must be with facts, not values. He argues that if it is accepted that some facts are already ‘valuey’ then Bhaskar's fact-to-value argument breaks down, because one would start with values, not facts, and consequently explanatory critiques would not be possible. In response to this, he argues that Bhaskar's value-to-fact argument is actually a fact-to-value argument. His case is that the statement about millions of people being massacred ‘[i]s both the most evaluative and the most precise and accurate; it gives more of the truth than the [other three statements]. That is so, but the evaluative force arises entirely out of the factual content’ (Collier, 1998, p. 451, emphasis added).

The problem for Collier's argument, though, is that the description of the Holocaust as a massacre of millions is normative – and appropriately so (as Berlin and Bhaskar argue). Accepting that this is an appropriately normative description does not mean that we have to accept Bhaskar's value-to-fact argument, though, and instead it can be presented as a value-to-value argument. For facts can be ‘valuey’ not because they have one set of agents’ values somehow ‘built in’, but because our prior value commitments shape how we describe some facts and what value judgements we make about those facts. In the case of the Holocaust it is described as a massacre of millions because this is how people's prior value commitments lead them to describe it and judge it. Describing the Holocaust as a massacre of millions applies values – it does not reflect values that reside in an object that has had agents' values impregnated into it.

None of the above criticism of the value-to-fact argument has shown that the argument is necessarily wrong. Instead it has just said that there may be instances where criticism needs to break from agential values and has rejected Collier's attempt to turn the value-to-fact argument into a fact-to-value argument. Below it is argued that the value-to-fact argument is necessarily wrong because it negates the need for people to make value judgements.

Rational value change

If we accepted Bhaskar's fact-to-value argument and/or his value-to-fact argument, then, for reasons of cognitive economy, we would have to be an ethical ‘blank sheet’. We would have to have no prior value commitments because such commitments would be redundant given that our values would be selected for us with: explanatory critiques forcing us, via logical derivation, to negative value judgements of certain structures and beliefs; and value-impregnated objects forcing us to one descriptively adequate account that could reflect the one set of values intrinsic to the object of study. There would be no rational basis for agents or political scientists to have their own value commitments which they would use to judge the political world because our values would just be determined by logically derived normative critiques of structural facts and/or accurate descriptions of ‘valuey’ facts impregnated with agents’ values.

Alternatively, we may say that values can be held and changed through a process of rational reflection. So, for example, someone may believe a war is justified and hold that the death caused is a ‘necessary evil’, only to revise this view later on when considering the human impact of the war and reconsidering its strategic objectives. While it is possible to be irrational (dogmatic, prejudiced, etc.) when it comes to values, it is also the case that people can hold their values through rational reflection on how good or not those values are in dealing with events.

In conclusion, the criticism of Bhaskar's arguments is not meant to motivate a return to the positivist commitment to value neutrality (which, ironically, is of course based on a value judgement to value value-neutrality over value-judgements). Rather, it motivates a critical approach to politics, with criticism rooted in rationally held values rather than putatively logical derivations or ‘valuey’ facts. The value-to-value argument gives values their due.