Understanding ‘Why’: How implicit questions shape explanation preferences

Abstract

‘Why’ questions are semantically ambiguous. A question like “Why is the sky blue?” can be rephrased as either a ‘how’ (“How did the sky get its blue color?”) or a ‘purpose’ question (“What is the purpose of the sky being blue?”). This semantic ambiguity allows us to seek many kinds of information with the same ‘why’ question. As a result, ‘why’ questions have often been used to investigate people’s explanation preferences. From such work, we know that people will often prefer teleological over mechanistic explanations—a tendency that has been linked to many broader theories of human cognition. But are ‘why’ questions pragmatically ambiguous? You may, for instance, have a specific expectation about what “Why is the sky blue?” was really meaning to ask. Here, we show that (a) people have clear, domain-specific expectations about what specific questions are implied by ambiguous ‘why’ questions; (b) people have clear preferences for certain kinds of questions over others; and (c) there is a direct link between implicit questions and explanation preferences. Thus not only is “why” pragmatically unambiguous, but these specific expectations may shape known explanation preferences. To test this view, we finally show that people endorse teleological answers even when they are explicitly non-explanatory. In other words, people may sometimes prefer teleological answers because they interpret ‘why’ questions as ‘purpose’ questions (rather than as ‘how’ questions) and teleological explanations may simply better address these questions. We discuss how understanding ‘why’ may reshape our understanding of people’s explanation preferences and their consequences.

A right way to explain? Function, mechanism, and the order of explanations

“…Prior studies have considered people's preferences for functional vs mechanistic explanations (e,g. Kelemen, 1999;Kelemen et al, 2013;Trouche et al, 2018;Chuey et al, 2020;Joo, Yousif, & Keil, 2021). In contrast, the current studies do not ask participants to favor one element over the other, but rather to consider how these two elements work together to constitute an explanation.…”

“…Prior studies suggest that childrenand even adults under cognitive loadendorse teleological explanations, often over mechanistic explanations (Joo, Yousif, & Keil, 2021;Kelemen, 1999). Studies 1a and 1b provide evidence that this preference for function might be due to a sensitivity to order.…”

“…Scientific phenomena can be explained at multiple levels (Craver, 2015). As a given phenomenon can be appropriately explained by appealing either to mechanism or to function (Joo, Yousif, & Keil, 2021;Lombrozo & Wilkenfeld, 2019), the existence of embedded levels of mechanism suggests the presence of levels of functions as well. For example, one might explain the function of an artifact as a whole (e.g.…”

Understanding ‘How’: Two kinds of mechanistic explanations underlie known explanation preferences

“…No other information was collected. For the sake of consistency and comparison with earlier experiments (see also Joo, Yousif, & Keil, 2020;Kelemen, 1999), data were analyzed in both through simple binomial tests and through t-tests treating participants' scored responses as averages. Redundant analyses are not reported.…”

Teleology beyond explanation

“…Thus, the claim that the mountain has a telos might not involve anything mystical or supernatural. It might simply reflect certain relatively mundane views about people's practices and how successful different objects are for those practices (see also Joo, Yousif, & Keil, 2021).…”