Like explanations, patterns are backward-looking, though they are also sensitive to present occurrences. Patterns can be used to predict future behaviour in relevantly similar circumstances to those in which we originally identified the pattern. We will embed the patterns we discover through investigation when we consider Requests for Prediction.

To use a pattern as a reason is to argue by analogy. To create good argument by analogy, we need to compare one situation to another. In doing so, we focus our attention on relevantly similar features of the circumstances. As always, this requires that we exercise our judgement.

For example, Freddy won the university bicycle race in Auckland last year. He also won the university bicycle race in Dunedin last year. And the one in Wellington. And the one in Christchurch. These were the only races he entered all year. Given these data, we could express any one of the following:

Freddy wins all the university bicycle races in New Zealand.

Freddy wins every bicycle race he enters.

Freddy always wins university bicycle races.

Freddy always wins when he races in New Zealand.

(...and so forth)

Our experiences in the world will influence our judgement about how we should characterise the pattern. Though all of the expressions above are reasonable expressions of the pattern, each is only an approximation of what might happen in the future. For example, Freddy might win every bicycle race he enters, but these four university races are probably not good predictors of success in the Tour de Monaco against professional racing teams.

Furthermore, our experiences in the world do much of the work in motivating us to seek patterns. On the one hand, we might ask “Why do we think there is a pattern to   X  ?” Or, we might ask “Why do we think that   X   follows a pattern?” Or, “Why does   X   happen with such regularity?” In any of these cases, and many others we could offer, our observation of some behaviour,   X  , gets us thinking that we should figure out what’s happening.

Often, when nothing stands out, the fact that nothing stands out, itself, stands out. We expect some variety in who wins the university bicycle races around New Zealand, and the fact that nobody other than Freddy has won stands out as novel.

Let’s work a generic example to illustrate the concepts:

LQ: “What happens to   X   in circumstances   Y  ?”
EE1: X enlarged in circumstance Y (at time T).
EE2: X enlarged in circumstance Y (at time T+1).
EE3: X enlarged in circumstance Y (at time T+2).
RA1: X always gets larger in circumstance Y.
RA2: X gets larger in circumstance Y when we measure its size.
RA3: X gets larger in circumstance Y, but randomly.
ER1: Objects the size of X are not known to be affected by measurement.
BE: In circumstances   Y  , X always gets larger.

(In case this example seems unlikely, consider two possible scenarios: We can imagine X to be a balloon inside a room whose atmospheric pressure is dropping. As the pressure in the room drops, the balloon would expand. Alternatively, we could imagine X to be a balloon sitting in the sun. As the balloon got hotter, it would expand.)

Rival Answers will be an observed change, plus a condition. Our Best Explanation will be the Rival Answer with the most reasonable condition, given the Evidence to Explain and any Explanatory Resources.

“X always gets larger” is the Best Explanation in the general case above because in every trial, X has gotten larger. We have no reasons to think otherwise at this point, though, of course, other explanatory resources could undermine this explanation. For example, RA1 would be undermined (and RA2 underwritten) if we found, contrary to ER1, that the act of measuring some material like X influences the measurement we read (odd as it might sound). Or in the hypothetical case of the expanding balloon, if we take measurement for a long enough period of time, the balloon will inevitably explode, which would alter our explanation to “X gets larger until it bursts.”

Generally, if the Investigation is about a single occurrence, the results will be an Explanation. If the Investigation is about a series of occurrences, the result will be a Pattern. Either of these can later stand as supporting resources in Requests for Prediction. We can express the distinction as follows.

Explanations express what happened.

Patterns express what happens.

Predictions express what will happen.

Last modified: Wednesday, 7 February 2018, 10:17 PM