Estructura del cuestionario

Many proff ered explanations in the life sciences, psychology and the social sciences refer to functions of entities. These sciences are concerned with com-plex organized systems, the components of which contribute to the working of the system (organisms, human minds, societies and so forth). A functional explanation typically accounts for the role or presence of a component item by citing its function in the system. Functional explanations are o en associated with teleology, the idea that there is goal-directedness in nature, because a functional description refers (implicitly or explicitly) to the goal of the system.

Teleology was central to Aristotelian philosophy, which regarded ‘fi nal causes’

as most important, because they appealed to the purpose of things for which they were designed. A er the Scientifi c Revolution, however, teleology came to be regarded as suspect in all natural sciences, as it did not fi t into the mechanical philosophy that had replaced the Aristotelian system. (In the human sciences, there is still room for teleology, because humans do have goals and intentions; see Section 5). Thus, the challenge for contemporary phi-losophers of science is to analyse functional explanation in such a way that it is compatible with modern science.

4.1 Early Views of Functional Explanation

Traditionally, philosophers assumed that functional explanations try to account for the presence of an item by showing how it contributes to the preservation or development of a system in which it occurs. The standard example: humans have a heart because it has the function of pumping blood, and pumping blood is required to stay alive. However, such explanations cannot be reformulated as DN explanations: from the fact that a functional item contributes to the preservation of the system, one cannot deduce that the item must exist, because there are always alternative ways to fulfi l the func-tion. For example, blood may be circulated by means of an artifi cial heart.

This led Hempel (1965, p. 313) to conclude that functional explanations are illegitimate. Instead, valid scientifi c explanations of the behaviour of orga-nized systems and their components will take the form of subsumption under laws, and will not diff er essentially from explanations in the physical sciences.

In the case of complex biological and social systems, these laws are typically laws of self-regulation, which specify feedback mechanisms. Nagel (1961,

pp. 403–4), by contrast, argued that legitimate functional explanations do exist in biology, if items are de facto indispensable for growth and reproduction.

The demands placed on functional explanations by Hempel and Nagel are quite strong: they claim that their goal is to explain the presence of functional items, and that they should do so deductively. No wonder, critics have argued, that most actual functional analyses failed on their criteria; these criteria are simply too demanding. Thus, Cummins (1975) suggests that the aim of functional analyses is to show how the item contributes to the functioning of the system as a whole. Cummins claims that functions are dispositions (or capacities), and as such, they require explanation. Rather than explaining functions by subsuming them under general laws, Cummins argues that the function of a system should be explained by analysing it into a number of other dispositions, the organized manifestation of which results in a manifestation the system’s function. This implies that we can only speak of the function of an item against the (implicit or explicit) background of a chosen analytical explanation. The choice for a particular explanation depends on the context.

Accordingly, an item may have diff erent functions in diff erent explanatory contexts. For example, while the heart is usually ascribed the function of a blood pump (in the context of explaining the circulatory system), the fact that it makes a noise may be functional, for example, in the context of medical diagnosis. This contradicts the traditional view, which qualifi es pumping as the ‘real’ function, while the noise it makes is a mere side-eff ect.

To avoid trivialization, Cummins (1975, p. 764) lists criteria for evaluating when analytical explanations are truly interesting: the capacities in terms of which the function is analysed should be less sophisticated and diff erent in type, and the organization of the resulting system should be relatively com-plex. In his later work, Cummins (2000) has applied his account to explana-tions in psychology and cognitive science, which are typically directed to capacities, such as the capacity to see depth, to learn and speak a language, and so forth. A further articulation and development of Cummins’ approach has been presented by Craver (2007, pp. 107–62), who argues that, at least in the life sciences and neuroscience, analytical explanations of functions are typically mechanistic explanations (see Craver and Kaplan, this volume). It is this idea, which off ers a prospect of integrating functional explanation and causal-mechanical explanation, to which we will now turn.

4.2 The ‘New Mechanist’ Approach to Functional Explanation

Salmon’s causal-mechanical account of explanation, discussed in Section 3.1, is modelled on explanatory practices in the physical sciences. In recent years, philosophers have developed mechanistic approaches to functional explana-tion that are inspired by the contemporary practices of the life sciences.

The core of these approaches is an analysis of mechanisms in the spirit of Cummins: a mechanism is an organized whole that, by virtue of the interaction of its parts, produces specifi c behaviour or performs a particular function. Most ‘new mechanists’ place themselves in the tradition of causal-mechanical explanation and regard their proposals as complementing, rather than replacing, Salmon’s account. In their infl uential paper ‘Thinking about mechanisms’, Machamer, Darden and Craver (2000) characterize mech-anisms as entities and activities organized so that they produce the phenome-non-to-be-explained. Phenomena are explained by describing the mechanisms that produce them. Their account is further developed in subsequent publica-tions; see, for example, Craver (2007) and Darden (2008). Slightly diff erent characterizations of mechanisms have been proposed by Glennan (2002), and Bechtel and Abrahamsen (2005), but all agree on the essential elements:

biological mechanisms exhibit complex organization and, by virtue of the organized interaction of their parts, produce specifi c functional behaviour.

A drawback of most mechanistic accounts is that they are descriptive and lack normative power. Craver (2007, p. 161) has a empted to remedy this defi ciency by providing normative requirements for satisfactory mechanistic explanations.

Mechanistic explanations are typically not purely linguistic, but also use diagrammatic representations (Bechtel and Abrahamsen 2005, pp. 427–30;

Machamer et al. 2000, pp. 8–9). Scientists prefer diagrams because they can directly convey the spatial organization of mechanisms. Diagrammatic representations of complex mechanisms have clear pragmatic advantages;

they are far more tractable than linguistic representations. Moreover, reason-ing about diagrams can be facilitated by simulation tools, such as scale models or computer models.

Mechanistic explanations are typically not constrained to a single level, but span diff erent levels: the system is described in terms of its lower-level parts.¹ But this does not imply that explanation is a reductive aff air. Rather, as Craver (2007, pp. 256–71) argues, mechanistic explanations involve ‘inter-level integration’ of hierarchically organized mechanisms into one coherent mecha-nism, which results in a description of nested networks of mechanisms that explain phenomena at diff erent levels. For example, a mechanistic description of the circulatory system in terms of the heart, blood vessels, the kidneys, the lungs and so forth explains its activity of delivering oxygen and nutrients to the body, while its parts (e.g. heart) can be mechanistically described in order to explain their respective activities (e.g. blood-pumping).

Remarkably, whereas much traditional philosophy of biology focuses almost exclusively on evolution, the new mechanists pay scant a ention to evolution, and one might indeed ask whether their approach fi ts evolution at all (Darden 2008, p. 967). On the other hand, it has been suggested that

the mechanistic approach can be applied to domains outside biology, such as chemistry and cognitive psychology (Bechtel 2008; Darden 2008; Ramsey 2008).²