Transitive inference and transitivity: Two sides of the same coin?

Lazareva (2025) read every paper that used transitive inference or transitivity tasks. She asked a simple question: do the two tasks measure the same thing?

The review covers animal labs, human labs, and brain studies. No new data were collected; she mapped what is known and what is still fuzzy.

The evidence is too thin to say the tasks share rules or brain paths. Papers often swap the names as if they are identical, but the procedures differ in small yet key ways.

Until studies directly compare both tasks in the same subjects, we should treat them as cousins, not twins.

Kao et al. (2024) extends the warning. They gave transitive inference to kids with autism and found the kids solved the puzzles just as well, only slower. The result shows the task works in a new group, but also that speed can mislead—exactly the kind of detail Lazareva says we must track.

Moustakis et al. (2018) used a stimulus-equivalence transitivity task and boosted scores by having learners invent a short story that linked the items. Their positive outcome looks like it clashes with Lazareva’s caution, but it doesn’t: they used a different procedure (equivalence training, not animal-style TI). The papers agree we need to name our method before we claim it teaches the same skill.

Hughes et al. (2014) came first. Their review said animal associative learning might mirror human stimulus equivalence. Lazareva (2025) narrows the lens to just TI versus transitivity, showing the field still hasn’t closed the gaps Sean flagged a decade earlier.

If you run or interpret derived-relations assessments, check the task label in your protocol. Swap the name only after you have data showing the two arrangements produce the same learner moves. Until then, write “transitive inference (five-item serial)” or “stimulus-equivalence transitivity” in your reports so future meta-analyses can sort apples from oranges.