Paul Schroder once observed that what really required an explanation was less why World War I broke out but rather why it did not break out in 1912 or 1913. This shift in explanatory purpose from “why did World War I break out at all” to “why did it break out in 1914 rather than in 1912 or 1913” reframes the question by focusing less on “why at all” and more on “why then.” This etiological shift makes sense if one is of the view that conditions conducive to a world war existed in the years prior to its actual outbreak.
This is an example of how a focus on explaining events neglects non-events that are in similar or even greater need of explanation. It is priors and expectations that make a non-event etiologically interesting. Unusually, the question why the Cold War never turned into World War III has garnered significantly more scholarly attention. Here again, an expectation, based on historical regularity, makes a non-event intellectually interesting. World War III should have broken, but it didn’t. Why? Standard answer: the existence of nuclear weapons and mutually assured destruction "did it."
Non-events garner less analytical attention than events and they attract less attention than they should. This may be because humans are cognitively wired to notice change and therefore have an innate tendency to want to explain changes rather than non-changes. Intuitively, a change can be accounted for by another change (or changes) preceding it, whether in terms of causation or correlation.
But this is a contestable presumption and assumption. Both an event (change) and a non-event (no change) can be picked out as the cause of another event (change). At least, this is true in terms of the counterfactual conception of causation. What caused the match to light? Was it because somebody struck it or was it because there was enough oxygen present? What caused a person to die of thirst? Was it the absence of water? Often “unchanged” factors are regarded as background conditions. This is legitimate as long as one admits that background conditions are part of a sufficient constellation of causal conditions. This does not always happen, thereby leading to a discounting of relevant background conditions as contributory causes.
Put differently, what if a match is struck repeatedly and it does not light until it is finally struck in the presence of oxygen? In this case, the oxygen would be widely seen as the cause? But is this really justified? Returning to the outbreak of World War One: the assassination of the archduke was equivalent to the striking of the match, the presence of oxygen was equivalent to the international political situation. Or was it the other way round? The point is this: Non-events can be conceptualized as causes of events (and non-events) under both a counterfactual and a regularity interpretation of causation.
What makes non-events as causes and effects interesting in the first place is the role they play in conceptual frameworks (theory) or as elements of an established (or posited) empirical regularity (history). In other words, inductive or deductive inference needs to assign etiological relevance to them. If both inductive and deductive inference suggest that an effect should be preceded by a specific cause, but the effect takes place without the expected cause being present, the non-event becomes intellectually interesting. In the Schroder example cited above, it is a non-event as the absence of an effect that is interesting because the presence of a cause (or causal constellation) should have led one to expect the effect to occur (World War I prior to 1914).
The focus on events, as opposed to non-events, also creates explanatory biases. In hindsight, historical events and sequences of events seem logical in the sense that they appear explanatorily coherent and (largely) consistent with the facts. But this can be deceiving, psychologically and epistemologically, precisely because the facts have been subsumed under some intuitively plausible model (or interpretation or explanation). But a sequence of historical events can be subsumed under a large number of different models, many of them fairly coherent and explanatorily satisfying. But re-interpreting or providing a different explanation for an event may also be constrained by the available evidence or lack of evidence. A plausible potential theoretical alternative may require evidence that exists but because it has not been entertained has not not been looked for. This is where researchers and historians enter the picture: they are the ones who will not or should not simply accept an explanation only because it is logically coherent and consistent with the available evidence. This may make for a good explanation, but it may be be faro from being the best possible explanation. Providing a better explanation requires offering an alternative or modified hypothesis and evidence that support it.
Sherlock Holmes realized that a non-event (the dog that didn’t bark) provided an explanation for a hypothesis that in turn allowed him to explain who committed the crime. The dog must have known the person that entered the stable at night, otherwise it would have barked. This led the detective straight to the criminal. Sherlock Holmes seems to have a keen appreciation of the importance of non-events. (A lot of what goes by intelligence analysis would benefit greatly from an appreciation of the potential importance of non-events as well as the potential relevance of the absence of evidence.) What makes this so interesting is that a non-event tied to a secondary hypothesis helped the detective solve the case. This is intellectual creativity.
The presence and absence of evidence in terms of confirming or disconfirming hypotheses is a long-standing issue in epistemology. Only because there is no evidence does not mean that the hypothesis this evidence is meant to support is not true. But such an absence is unlikely to make the hypothesis likelier to be true. The absence of evidence can matter to the extent that one has legitimate expectations as to whether the evidence should be found or not. This of course requires one to posit auxiliary hypothesis and make assumptions as to how likely should be to find evidence provided it exists. It also matters where one looks for evidence and how much effort one expands searching for evidence (aka “streetlight effect”). So sometimes the absence of evidence may provide some evidence for its absence, and hence evidence for or against a given hypothesis. But not always.
Anthropologists have trouble finding evidence for humans having inhabited certain areas of the world. This may be because the climatic conditions over time have made it unlikely that any evidence remains. But anthropologists might expect, based on their otherwise well-supported model or explanation, to find little evidence of human habitation in another part of the world because their model suggests that very few, if any humans lived there. But because evidence was more likely survive in these areas, researchers would be wrong to conclude that humans were more plentiful in areas where evidence is found, compared to areas where no such evidence is found (despite significant efforts to find it). In other words, evidence (and its absence) bears more or less strongly on a given hypothesis, depending the assumption one makes with regard to other plausible, but not necessarily strongly confirmed auxiliary hypotheses. This is where Bayesianism enters the picture!
Non-events and the absence of evidence are very important epistemological concepts. Their explanatory and confirmatory relevance is on a par with events and evidence.