Dogs outperform cats both in their testability and relying on human pointing gestures: a comparative study

Why it matters—and what it asks

Can cats use a human’s pointing gesture to locate food, and how does their performance compare with dogs? Crucially, the study separates testability—whether an animal makes a choice—from accuracy when choosing. No earlier reading required.

Method, in plain language

An experimenter pointed toward one of two bowls, with her fingertip about 80 cm from the target. She either kept pointing until the animal chose or withdrew her hand before its release. The protocol offered 28 trials.

Laboratory choice analyses included 33 cats and 21 small dogs. Accuracy analyses retained animals choosing in at least half the trials: 15 cats and all 21 dogs. A subset of cats was also tested at home. Researchers compared counting no-choice trials as failures with excluding them.

Key findings

• Dogs made more choices and were more accurate than cats in the laboratory, under either scoring approach.
• Cats exceeded chance when no-choice trials were excluded, but not when counted as failures. That changes the question from overall task success to accuracy conditional on participating.
• Cats made more choices at home; their participation declined across laboratory trials but not home trials. There was no clear accuracy advantage for either pointing style.

Important limitation: Cats were heavily preselected for tolerating the laboratory, and home testing always came later. Stress, motivation, handling and experience cannot be cleanly separated from communicative ability. This is not evidence that cats lack gesture comprehension—or that dogs are generally “smarter.”

Reading route

Abstract → Table 2 → Procedure → Figure 4 caption and both scoring analyses → Discussion.

Two questions:

1. When should not choosing count as failure, and when should it be analyzed separately?
2. How could we make a cat–dog comparison fair without assuming identical conditions feel equally comfortable?

Guided reading — Lumi

Explanatory synthesis of the existing verified note, not new research. The earlier PMC full-text audit covered substantive methods and results; figure captions, not pixels, were read, and supplements were not independently audited.

1. Distinguish the two pointing cues

• What to look for: An experimenter indicated one of two bowls, either maintaining the point until a choice or withdrawing it before release. The fingertip was away from the target bowl.
• What it means: Sustained pointing leaves a visible cue available; momentary pointing removes it before the animal chooses. Both test performance using a human directional cue.
• Why it matters: The absence of a clear accuracy difference between gesture types is not proof that the cues are equivalent, or that animals interpret human intentions.

2. Separate participation from accuracy

• What to look for: Researchers analyzed whether animals chose at all, then accuracy among animals meeting a minimum-choice threshold. They also compared treating no-choice trials as failures with excluding them.
• What it means: Testability measures participation in this setup. Accuracy conditional on choosing asks a narrower question than success across every offered trial.
• Why it matters: Cats exceeded chance only when no-choices were excluded. Refusing or failing to choose is not the same observation as selecting the wrong bowl, although either can count against practical task completion.

3. Read the species comparison without ranking intelligence

• What to look for: Dogs made more choices and were more accurate under both scoring approaches. Only a selected subset of laboratory cats entered the accuracy analysis.
• What it means: The species difference is real within the analyzed task, but depends on animals’ engagement as well as choice performance.
• Why it matters: Laboratory tolerance, motivation and handling can affect participation. This is not a general intelligence test or evidence that every cat lacks gesture comprehension.

4. Examine the home comparison’s limits

• What to look for: Cats made more choices at home, but home testing always followed laboratory testing; dogs were not tested at home.
• What it means: Location and prior experience changed together, preventing clean separation of their effects.
• Why it matters: A reader might propose balancing location order in a future study. That is a hypothetical improvement, not a control used here, and greater participation alone does not establish improved understanding.

Additional interpretation notes — Lumi

The 43 cats arriving for testing were already drawn from a prior habituation process. The choice analysis used 33 laboratory cats; only 15 met the at-least-14-choices threshold for success analysis. At home, 23 arrived, 19 contributed choice data and 14 met the success threshold. These are distinct denominators. Only three cats met that threshold at both locations, so strong within-cat accuracy conclusions are not available. Dogs were tested in the laboratory, not at home; the discussion contains a loose reference to dogs in both locations that should not be repeated. Procedure also gives 22 dogs in its break description, whereas Subjects, Table 2 and Results consistently give 21; the card follows the convergent analyzed sample.

Both scoring approaches show a species difference. The above-chance cat result is conditional on a selected subset and, when no-choices are excluded, on actual choices. It is not evidence that every cat follows pointing reliably. No demonstrated difference between gesture types is not proof of equivalence. Older cats were more successful, but age was not randomly assigned. The authors’ evolutionary explanations are interpretations, not experimental isolation of domestication effects. Neither nonparticipation nor success establishes understanding of human intentions.

Connections

• Investigates: Human pointing gestures — sustained and momentary distal pointing.
• Uses method: Testability and no-choice trials — distinguishes participation from choice accuracy.
• Compares with: 2021 Zhang - Unsolvable task and human attention — both involve food-based social tasks with selected cat samples, but they measure different behaviors.

Sources

• https://www.nature.com/articles/s41598-023-45008-3
• https://pmc.ncbi.nlm.nih.gov/articles/PMC10587310/
• https://api.crossref.org/works/10.1038/s41598-023-45008-3

Source access and verification scope

Verified 2026-09-23 Asia/Seoul. Original Nature article establishes title, journal, article number and open-access status. PMC substantive Methods, Results, Discussion and license inspected, including the extracted middle Results section. Crossref independently checked all five authors, exact title, volume, issue, article number and online date; no separate print date supplied. DOI-specific correction/erratum search surfaced no result; inspected sources showed no correction notice and Crossref update-to was absent. This bounded check does not guarantee absence of later corrections. Figure captions were read, not figure pixels; supplements were not independently audited.