Entity realism and experimental failure

Entity realism, defended most famously by Ian Hacking, is the view that we should be sceptical of scientific theories, but we should believe in unobservable entities that scientists can manipulate and use as tools to study other phenomena. Nobody will ever observe an electron, but we should believe in electrons because we can spray them onto deuterium to help study weak neutral currents. As Hacking says, "When we use entities as tools, as instruments of inquiry, we are entitled to regard them as real."

Discussions of entity realism often assume that experimental or manipulative success is the entity realist's criterion for belief in unobservable entities; Gelfert, for example, says: "In its original form due to Ian Hacking, entity realism postulates a criterion of manipulative success which replaces explanatory virtue as the criterion of justified scientific belief." I suggest that the entity realist would be better off dropping this criterion. Experimental failure often provides just as strong reason for belief as experimental success.

Consider the following case from solar astronomy. Thermonuclear fusion in the core of the Sun occurs by the proton-proton chain, where four protons are converted into a helium nucleus, two positrons, and two electron neutrinos:

4p → ⁴He + 2e⁺ + 2ν_e

Neutrinos interact with matter extremely rarely; a piece of lead one light-year thick would stop only 50% of any neutrinos passing through it. The vast majority of neutrinos produced in thermonuclear fusion pass straight through the Sun unimpeded. So studying neutrino flux gives us a direct view into the core of the Sun. The challenge is how to detect the flux of these particles that interact so weakly with matter.

The first experiment to detect solar neutrinos was developed in the mid-60s by Ray Davis using a tank of perchloroethylene, C2Cl4. Very rarely, a neutrino collided with a chlorine nucleus, converting it into radioactive argon by converting a neutron into a proton. The radioactive argon was then captured by bubbling helium through the tank. Davis used the rate of argon production to measure the neutrino flux. Unfortunately, the Davis experiment was a failure; he detected only 1/4 of predicted neutrino flux (figures are from Longair's The Cosmic Century, p.180):

Predicted flux: 7.9 +/- 2.6 solar neutrino units
Observed flux: 2.1 +/- 0.9 solar neutrino units

What went wrong? Different reactions in the p-p chain produce electron neutrinos of different energies. It was proposed that Davis was detecting only the high-energy neutrinos. Further experiments, such as GALLEX and SAGE, which were based on neutrino collisions converting gallium into radioactive germanium, were designed to detect the low-energy neutrinos. These experiments also failed; they detected only about half of the predicted low-energy neutrino flux. Another suggestion was that something was wrong with the solar models – perhaps we were mistaken about the temperature in the core of the Sun, for example – however, by the 90s experiments in helioseismology were providing robust confirmation of these models, and it became clear that the problem lay with our understanding of the nuclear physics relating to neutrinos (see Longair p.182).

The problem was finally resolved in the early 2000s by the discovery of neutrino oscillation: some of the electron neutrinos produced in the p-p chain spontaneously converted into tau and muon neutrinos, which were not detected by any of the original experiments. However, what is interesting in the context of entity realism is the situation in the decades prior to this. From the mid-60s to the early-00s, across numerous different experiments, predictions failed and experiments did not work as intended. So any argument from experimental success could not apply to neutrinos. Yet it remains the case that scientists used neutrinos as tools, as "instruments of inquiry", in their investigations of the core of the Sun, which seems to be what Hacking requires to justify belief in neutrinos. Indeed, few, if any, scientists working on solar physics doubted that neutrinos were being detected.

This case suggests that there is nothing especially important about experimental success. By analogy, we use hammers to drive nails into walls. There are many ways we might fail: we might smash a hole in the wall, or we might be using a dud nail that snaps when we hit it. None of this would lead us to doubt the existence of the hammer. We don't infer the hammer's existence from its success at driving nails into walls; the fact that we can use and manipulate the hammer in certain ways is enough.

One reason why it's worth exploring cases of experimental failure is that these may provide the entity realist with the resources to answer the persistent objection that entity realism collapses into standard scientific realism. Believing in some entity X, so the objection goes, requires believing theories about X. As Alan Musgrave puts it:

To believe in an entity, while believing nothing else about that entity, is to believe nothing or next to nothing. I tell you that I believe in hobgoblins. "So", you say, "You think there are little people who creep into houses at night and do the housework." To which I reply that I do not believe that, or anything else about what hobgoblins do or what they are like - I just believe in them.

Hacking, as is well known, has claimed that while we shouldn't believe scientific theories about e.g. electrons, there are basic "home truths" about electrons, a basic "common lore" about electrons, that we should all accept. Such home truths include that electrons have negative charge, that they "orbit" (in some sense) atomic nuclei, that they have a rest mass of about 9.1x10-31kg, that certain materials transmit electrons more readily than others, etc. This is enough to give us a substantial belief in electrons, and to allow us to use electrons in scientific experiments.

Theories change but home truths remain. This is illustrated very nicely by the neutrino experiments. Since the neutrino detection experiments failed to confirm the theoretical predictions, it was clear that some part of the theory was wrong. Initially this was attributed to inaccuracies in the solar models, but by the 1990s most scientists turned their attention to the nuclear physics. Neutrino theory was wrong. Arguably then, belief in neutrinos could not be justified by appealing to this theory. Nevertheless, scientists retained certain low-level assumptions or "home truths" about the causal powers of neutrinos, such as that neutrinos are produced by reactions in the p-p chain, and that they can very occasionally collide with chlorine converting it into argon. Perhaps entity realists should turn their attention to examples of experimental failure.

An interesting follow-up question for the entity realist would be, how do we distinguish cases of failed prediction where we nevertheless assume that the experimenters were using a certain unobservable, as occurred in neutrino astronomy, from those cases where we think that nothing important was detected? Joseph Weber's research into gravitational waves might be an example of the latter case.