Do 17-Month-Olds Really Know “If Not A, Then B”?

Can a baby who has not even learned to speak yet perform a logical deduction like, “If it is not A, then it must be B”? A study by Feiman, Mody, Sanborn, and Carey (2022) suggests a fascinating possibility: between approximately 17 and 19 months of age, infants begin to use negation to exclude an invalid option. However, looking more closely at the experimental evidence raises an important question: when an infant behaves as though it has eliminated one possibility, is the infant actually performing reasoning by exclusion, or could the behavior be produced by a simpler cognitive strategy?

The most immediate possibility that comes to mind is a “novelty preference”. In the blicket detector task, infants first see blocks A and B activate the machine together, and then see that A alone fails to activate it. When given a choice between the two blocks, a 19-month-old will often choose B. But is this genuinely reasoning by exclusion—“A does not work, therefore B remains a candidate”—or could the infant simply be attracted to the block that has not yet been tested?

To distinguish these explanations, I would propose a three-block control experiment. After the original evidence is presented, a third, completely unmanipulated block C could be introduced alongside A and B. If the infant’s choice is driven primarily by novelty, the infant should be attracted to C, which has received even less prior manipulation. If the infant is actually using exclusion, however, the infant should specifically select B, because B is the remaining candidate after A has been ruled out. Such a result would provide stronger evidence that the infant is responding to the logical structure of the evidence rather than simply to novelty.

A second concern is whether the failure of younger infants necessarily reflects the absence of the relevant logical capacity. A 15-month-old may understand that one cup is empty yet still fail to use that information to guide an action. In fact, Feiman et al. explicitly investigate several possible performance limitations. Experiment 2 provides evidence that 17-month-olds can have difficulty disengaging their attention from the bucket that was most recently manipulated.

Experiment 3 then tests alternative explanations for 15-month-olds’ failure, including insufficient working memory, insufficient motivation, perseveration, and difficulty inhibiting attention to the last-manipulated item. The 15-month-olds nevertheless succeed when they do not need to exclude an option in order to find the toy. The authors therefore argue that the failure is more consistent with an inability to use the information that one bucket is empty to exclude it from consideration, rather than simply with a general inability to perform the task.

The developmental pattern becomes particularly interesting when the two experimental paradigms are compared. In the 2-cup location task, 17-month-olds approached the correct bucket on 69% of trials, significantly above chance. In the blicket detector task, 17-month-olds chose the causally relevant block on 60% of exclusion trials, also above chance.

Although the two percentages should not be treated as a direct measure of the relative difficulty of the two abilities, the difference raises an interesting question about how the demands of the two tasks interact with infants’ developing reasoning abilities. The location task requires infants to reason about where an object is, whereas the blicket detector task requires them to reason about an invisible causal property of objects. Importantly, Feiman et al. emphasize that the two paradigms differ in the content being reasoned about, while showing a similar developmental trajectory. This convergence is therefore consistent with the possibility that infants are developing a content-independent capacity for exclusion.

Of course, the logical ability demonstrated at this age should not be assumed to be equivalent to fully developed disjunctive reasoning. Reasoning by exclusion can be represented as the exclusion component of a disjunctive syllogism: if P or Q, and not P, then Q.

Feiman et al. distinguish this emerging ability to use negation from the later ability to represent the disjunctive relation itself in a fully developed way. Their discussion considers the possibility that reasoning develops in two steps: exclusion based on negation may emerge around 17–19 months, while relating multiple alternatives through a disjunctive premise may develop later, around age four.

This distinction makes the findings more interesting rather than less impressive. A 17-month-old may not yet possess the full logical machinery of an adult, but may nevertheless be beginning to use negation as a tool for eliminating possibilities. The central question, then, is not simply whether infants can “do logic,” but what kind of logical representation their behavior reflects.

The fact that exclusion emerges at a similar developmental period across spatial and causal tasks suggests the possibility of a content-independent logical capacity, but further experiments are needed to determine whether this capacity is genuinely shared across domains or whether apparently similar behaviors can arise from different underlying representations.