Department of Philosophy, University of Auckland.
Throughout The Logic of Scientific Discovery (LSD) Popper proposes a number of rules which, he thinks, ought to govern the scientific enterprise and which comprise his theory of scientific method. Some of these rules have been extensively debated. Hardly discussed is a rule in which Popper proposes to eliminate all observations which do not arise in the course of the testing of a theory. I wish to argue that such a rule ought to have no place in any theory of method because it would exclude much that is valuable in science.
In the LSD 30 Popper claims that `basic statements are accepted as the result of a decision or agreement...' and that such `decisions are reached in accordance with a procedure governed by rules'. In fact Popper gives us no full set of rules for the acceptance of basic statements, but he does mention two such rules. First there is a rule which tells us what is a basic statement. This is set out in LSD 28, the gist of which is that basic statements have the form of singular existential statements (which are restricted in scope), i.e., they are of the form: (x)[(x is in a finite space time region R). (x is )], where `' refers to an observable event process or property (`observable' is left undefined but is explained via examples). Second, there is the rule (LSD 30):
Of special importance amongst these [i.e., the rules in accordance with which decisions are made to accept or reject basic statements] is a rule which tells us that we should not accept stray basic statements - i.e., logically disconnected ones - but that we should accept basic statements in the course of testing theories; of raising searching questions about theories, to be answered by the acceptance of basic statements.
The rule, as stated, does not fully capture Popper's intentions since he subsequently talks not only of accepting basic statements but also of rejecting them. More explicitly, the second basic statement rule (BSR) can be expressed as follows: BSR:(i) Do not accept stray basic statements; (ii) only accept or reject basic statements in the course of testing a theory.
In criticizing BSR I wish to set aside objections to the "decisionist" or "conventionalist" view Popper espouses concerning basic statements (see the latter parts of LSD 30) which comprises his anti-foundationalist epistemology. Rather, I wish to concentrate on the worth of BSR as a rule of scientific method. True, when we test theories we do, in the course of the testing, accept or reject basic statements, and we ought to do so. But is the converse the case? One version of the converse can be expressed factually: when we (or scientists) accept or reject basic statements then we do so only in the course of testing theories. This factual claim is false, one counter-example being Popper's naive inductivist who, while collecting basic statements, has no particular theory in mind, or has no particular theory under test at the time of the collecting. However BSR is not factual but normative. So a second version of the converse is prescriptive: one ought only to accept or reject basic statements in the course of testing a theory. It is this normative form of BSR I wish to criticize.
Before doing so there is a further point of unclarity about BSR. Since basic statements are falsifiable then on Popper's demarcation criterion they should at least be accepted as scientific and not rejected as unscientific. Does this mean that BSR is in conflict with the demarcation rule in excluding a class of statements the latter would include within the realm of science? If BSR has priority over the demarcation rule then we would lack the entire class of basic statements whereby the demarcation rule is possible. But perhaps this objection is mistaken because there is a mistake about the purpose of acceptance and rejection in BSR. Acceptance and rejection in BSR may not be for the purpose of classifying statements as scientific or non-scientific. Perhaps acceptance and rejection in BSR are for the purpose of testing theories. If so, BSR is a uninteresting analytic truth about basic statements. More likely Popper intends acceptance and rejection in BSR for the purpose of assigning a truth-value to basic statements; that is, we are to accept a basic statement as true or reject a basic statement as false only in the course of testing a theory, and not otherwise. Such an interpretation fits better with Popper's anti-justificationism for basic statements. We can accept them as true or reject them as false recognizing that a different verdict may be possible for each basic statement since none of them is incorrigible or certain.
With this point now made we can ask: should scientists only accept basic statements as true, or reject them as false, while in the course of testing a theory? I will answer `NO!' Scientists should be permitted to accept a basic statement as true or to reject it as false while not currently testing any particular theory. The reason is that the stock of significant observations made in science would be seriously depleted and science thereby rendered less progressive than it has been. I will cite four examples of observational reports which can be expressed in the form of basic statements, which are important observations, which are `stray' in the sense of BSR and which have been, and ought to have been, accepted without these statements arising in the course of testing a theory.
Example 1: As is well known, Alexander Fleming noticed that one of his culture plates contained staphylococcal colonies which were undergoing dissolution near a foreign contaminating mould. Being an untidy worker Fleming did not throw the plate away but kept it as a curiosity and subsequently transferred bits of the contaminating mould to other culture plates. Though Fleming was working on a number of problems that led him to make the culture plates in the first place, none of this bears on his observing what was happening on the contaminated plate. Fleming did have an interest, going back to the First World War, in the problem of infected wounds and, in particular, of whether or not there were non-toxic antibacterial agents; and perhaps this interest lead him to take a closer look on that fateful day at what is a common occurrence with culture plates, viz., contaminating mould. Certainly Fleming was not testing any particular theory at the time he made his report. His attention was simply grabbed by what had occurred on one of his culture plates. Should Fleming have acted in accordance with BSR? We would all be the poorer if he had.
Some might suggest that BSR be understood differently. Basic statements are not to be only accepted or rejected by scientists in the course of their testing a theory; they are to be accepted or rejected according to whether scientists entertain some belief or hypothesis to which they are logically connected, or the scientists have some expectation, conscious or unconscious, with which they are logically connected. Clearly this is very different from BSR and involves considerable broadening of the conditions under which basic statements are to be accepted or rejected. However it does accordwith claims Popper makes elsewhere about theory-driven observations in contrast with those philosophers who espouse a more empiricist view of observation.
There are two objections to this weakened version of BSR. First, for any basic statement that might be accepted or rejected there is always some hypothesis that could be gerrymandered as the object of belief of a scientist or, more weakly, as the conscious or unconscious expectation of a scientist. In such a weakened form BSR would be trivially true but hardly a claim worthy of attention. Second, and more significantly, to have a genuinely informative version of the weakened BSR in the case of Fleming we would have to show that he ought not to have accepted his observation report (expressed in basic statement form) unless Fleming actually did have a belief, or a conscious or unconscious expectation, to the effect that, say, contaminating mould do not normally kill off staphylococci. But this is unacceptable. As a matter of fact Fleming held no such belief, nor had any such expectation, since such contaminating mould were often found in the culture dishes of experimental scientists, Fleming included. Nor ought Fleming to have held any such belief, nor ought he had to have any such expectation, in order to accept his observational report of what was happening on the contaminated plate. There is simply no ground for such a prescription.
Example 2: In 1822 Dr. William Beaumont was lucky enough to come across the unfortunate Alexis St. Martin who had been shot in the abdomen. He had survived but his wound had not healed and left a permanent hole in the region of his stomach. Since no one had previously observed the processes of digestion the curious Dr. Beaumont struck up a relationship with the hapless St. Martin that enabled him to frequently peer into the wound to see what was going on. Here is a sample of Beaumont's observations: `At 9 o'clock he breakfasted on bread, sausages and coffee, and kept exercising. 11 o'clock, 30 minutes, stomach two-thirds empty, aspects of weather similar, thermometer 29 o [F], temperature of stomach 101 1/2 o and 100 3/4 o . The appearance of contraction and dilation and alternate piston motions were distinctly observed at this examination. 12 o'clock, 20 minutes, stomach empty.' Beaumont's observations give us the first account of what actually goes on inside a person's stomach, something of considerable interest to medical science. Clearly there is no theory under test here! Of course Beaumont did perform some experiments such as extracting stomach juices from St. Martin to test various hypotheses about how the juices perform their action on food. But none of these experiments bear on Beaumont's important observational record of what went on inside St. Martin's stomach from the time food entered until it left. On Popper's BSR the observation record of the actual digestive processes in the stomach would be excluded from science.
Example 3: For a host of reasons meteorologists have collected daily rain data and the maximum and minimum daily temperature, in some cases for several centuries. The recording of such data is clearly independent of any theory of the weather scientists may hold, especially the testing of some particular weather theory. However such data can reveal trends, for example whether or not the "greenhouse effect" is occurring. The "greenhouse effect" is an hypothesis about the heating up of the earth due to increasing amounts of carbon dioxide and other gases in the atmosphere. Meteorologists might well now be gathering temperature data to test the "greenhouse" hypothesis and so their acceptance and rejection of basic statements about temperatures takes place in the course of the test of a theory. But such data need not be collected only in the course of such a test and clearly was not when the "greenhouse" hypothesis had not even beenformulated. Importantly, early data collected before the formulation of the hypothesis is necessary for its test, both now and subsequently.
Example 4: Robert Hooke's Micrographia (1665) records the delight of looking for the first time at the features of small objects under a microscope. Quite untrammelled by the need to test any theory Hooke both records the observations he made and drew diagrams of what he saw, e.g., the rough point of a pin, the hairs on the legs of lice, the pores in cork, the various crystalline shapes of flakes of snow, and so on. The point of making such observations, carefully and scrupulously recorded by Hooke, was to expand the range of observable phenomena hitherto available to us and not to test any particular theory. Of course, any of these observations could be used by anyone else at any subsequent time to test some theory, but this is not necessary for either the activity of observing or the significance of the observation made.
What the above examples show is that the collection of data can be a significant scientific activity which can, and should, on many occasions take place in the absence of any theory contemporaneously under test by the collectors of the data. This is not to say that data collectors may not have some interest in collecting the data. Rather that interest need not be an interest entirely driven by the need to test some theory as BSR requires. Scientists can have other equally legitimate interests in collecting basic statements, i.e., observational reports, besides theory testing; e.g., sheer curiosity can be one impetus for making observations. It is these interests that bestow significance on the reports. The interest a scientist has in testing a theory would be sufficient to bestow significance on an observation; but no one particular interest is always necessary. Observations can become significant in science for a host of reasons having nothing to do with theory testing.