The Parallelism with Ordinary Experience

Empiricists philosophers are quite right in stressing the epistemic role of obser­vation or perception as warrant of our assertions of existence. In order to justify the belief in the existence of objects inaccessible to our immediate perception, we are only allowed to resort to arguments analogous to the ones used to comfort our beliefs in the existence of things immediately perceptible in the context of our everyday experience.

In the case of direct perception, it is evident that the first requirement is also satisfied, because the relevant properties are observable and perceptible. The second requirement is not, since in ordinary experience, it is not necessary to perform quantitative measurements to confirm the concordance between perceived proper­ties in different circumstances, in accordance with the fourth requirement.

The third requirement is also satisfied since it is easy to convince oneself that causal processes are at work in perception by verifying the systematic variation of some perceived properties in function of the changes in the observing conditions. Also, we can see that these properties vanish when the thing is removed. This is just a crude application of Mill's methods. Full knowledge of the laws and the causal mechanisms at work in perception is not indispensable because the immediate perceptual contact doesn't necessitate to resort to causal laws to guarantee that the perceived thing exists.

How do we infer the existence of a perceptible thing, which is not actually present in perception? In a by now famous example, van Fraassen (1980, 19-20) discusses the clues which permit to conclude that a—unobserved—mouse has come to live with me. Nobody has seen the mouse (which is observable...) yet, but some gray hair lies on the ground, cheese disappeared, tiny noises have been heard etc. Such evidence makes me believe that a mouse is living in the house. How is such belief justified? Well, we have seen mice before and we know that they have grey hair etc. We suspect that several causal links obtain between the existence of a mouse and the observed clues. Each causal relation points toward a particular property, such has loosing grey hair or eating cheese. The set of these properties and the awareness of the context allow to identify the thing which is causally responsible for their presence.

At this point, we come to grips with the justification of the belief in the existence of unobserved objects which are typically postulated by a theory to causally explain why some phenomena occur. The existence of molecules was asserted by Boltz­mann because he wanted to causally explain the properties of gazes, such as pressure, temperature etc. I distinguished above between two kinds of properties that can fulfill an explanatory role, the OP properties which are observable in principle and the PT properties which are purely theoretical and escape de dicto any possibility of observation, even by means of instruments. With respect to objects which possess PT properties only, an agnostic attitude is to be recommended until perhaps some theoretical investigations lead us to attribute OP properties to them.

Thus, the selective realism I defend can give credence to observable properties only, but these properties are not limited to directly observable ones: they include OP properties, that is properties which could be observed through adequate instruments. OP properties are eligible candidates for membership in our ontology. As regards the sense of sight, which enjoys a privileged status in the sciences, observation instruments are extremely varied. First, we have the banal glasses commonly used by short-sighted, farsighted and presbyopic people whose eyesight is mildly deficient. Who will dare to claim that the observations made by someone who uses appropriate glasses are less reliable because they were not performed directly with the naked eye? Next,^[46] we have the different kinds of microscopes and telescopes which make visible entities such as viruses, remote stars and galaxies.

Moving further away from ordinary perceptual experience, we can mention the cause of the light rays directly observed in the cathode ray tubes. These light rays are produced by the interaction of unobservable particles—the electrons—emitted by the cathode, which interact with the molecules of the rarefied gas filling the tube.^[47] J.J. Thomson succeeded in determining the values of the charge and the mass of an electron without knowing how the cathode rays were produced. How­ever, knowing the laws permit to manufacture instruments, such as wire and drift chambers to detect and measure the properties of charged particles. These devices make visible the trajectories of charged elementary particles, such as protons. In order to be justified in claiming that the lines immediately observed on photographs or computer screens represent the trajectories of protons, we must know the causal mechanisms, i.e. the causal laws^[48] which describe the interactions between the protons and the molecules of a gas such as argon. Typically, the protons remove electrons from the molecules of the gas. These electrons are instantly captured by the wires in their vicinity, which detect them and make visible the trajectories of protons.

This case is analogous to the emulsions studied by Perrin. The molecular agi­tation is rendered visible by the movements of the grains of mastic which collide with the molecules of the liquid, in the same way as the trajectory of a proton is made visible by the electric discharges resulting from the interactions of the protons with the molecules of the gas.

Whether we look at directly observable entities, like a rose or a mouse, or indirectly observable, such as the grains of an emulsion or galaxies, or at entities whose existence can only be inferred, such as atoms, electrons, protons or gravi­tational waves, our reasoning is grounded on the attested presence of causal rela­tions (how we attest such presence is examined in Sect. 6). The working of causal mechanisms, the laws of which can be known or not, is crucial to justify our belief in the existence of entities, whether directly or indirectly observable, or inferred only. Surely, actual perceptual presence in adequate conditions is what guarantees the truth of the assertion that an entity possesses directly observable properties. But perception is supported by causal mechanisms that we have reasons to believe to be present even if we might still ignore the details of their action, if only because when we intervene on an object in a certain way, systematic variations in the perception of its properties occur. In the same way, if we have a complete knowledge of the functioning of a measuring instrument, we have reasons to believe that the causes of certain observed effects are objects with specific OP properties. As a consequence, nothing prevents that these objects might be (indirectly) observed someday. Molecules can now be seen through instruments which were not at Perrin's dis­posal: the electronic microscopes.

My vindication of selective—local—realism is grounded on a parallelism between the reasons to believe in the existence of an ordinary thing and the reasons why we are rationally entitled to assert the reality of some concrete unobservable object and some true statements about it. Most scientists resort to such arguments to sustain the existence of objects which have properties which are de facto—but not de dicto—unobservable, namely the OP properties.

The concordance of the results obtained by the various independent methods of measuring Avogadro's number is comparable to the concordance of the various ways of observing the same property of a mouse. One could say that the finiteness of Avogadro's number is the best explanation of an obtained value. The same holds for the other finite values obtained by other methods. It is this concordance that certifies that the measured values are finite and approximately correct. In an anal­ogous manner, several perceptual accesses to a property of a thing, whether seen, touched etc., can attest that a perceived thing such as a tabletop is rectangular. The presence of the rectangular form explains, via various causal processes, the agreement between the various perceptions. But we know that each individual perception is immediate evidence that the tabletop is rectangular, irrespective of a possible knowledge the underlying causal processes. In the same way, when we deal with OP properties, whether of mice or electrons, we must have reasons to suppose that causal processes are at work and that they furnish the basis for the causal explanations of our direct or indirect observations.

For each theory and each posited object, we must examine if causal chains connect specific phenomena to some specific OP properties and if independent measuring operations generate concordant results for the same property. Every existence assertion deserves a special justification, as we saw for the determination of the Avogadro number and the masses and dimensions of molecules. Yet, the philosopher aims at universality. For this reason, I tried to show that each particular argumentation articulated by the realist must obey the four requirements that I formulated and motivated above.

6