Follow my look: attention and choice are different outcomes

Citation and access

Péter Pongrácz, Julianna Szulamit Szapu & Tamás Faragó (2019). Cats (Felis silvestris catus) read human gaze for referential information. Intelligence, 74, 43–52. DOI: 10.1016/j.intell.2018.11.001.

The final issue is May–June 2019. The authors’ university announcement dated November 16, 2018 cites the already available paper as 2018. Thus online year 2018 and issue year 2019 refer to the same study. The exact online day was not verified from an original bibliographic source and is deliberately left unset; a secondary search result is not sufficient to establish it.

Access: partial text, not verified free full text. The original publisher preview supplied the complete abstract and introduction plus truncated Subjects, Success, Discussion and Conclusions sections. The university’s explanatory account was also inspected. Crossref confirmed the exact title, all three authors, journal, volume, pages, DOI and May 2019 issue metadata. Full procedures, figures, tables, exclusions and supplementary materials were not available for a complete audit. An author-associated Academia page yielded only excerpts; its direct retrieval returned HTTP 403. A repository search did not establish a legal free complete manuscript. No correction was identified in the inspected publisher material or Crossref relation/update fields; the targeted correction search returned unrelated results, so this remains a bounded check, not proof of absence.

Why this paper matters — Lumi

Looking at a person and following that person’s look are different skills. This experiment asks whether a cat can use a human’s head-and-gaze direction to locate hidden food. Its especially useful lesson is that getting attention faster need not improve the eventual choice. This is an approachable companion to work on human attention and pointing, without requiring any previous reading.

Question and methods — Lumi

Can companion cats use a previously unfamiliar human’s gaze cue to choose between two locations, and does a communicative attention-getter help?

The publisher reports 41 companion cats tested in their owners’ homes. The Subjects preview says recruitment requested strongly food-motivated cats. According to the university account, an unfamiliar kneeling experimenter looked toward the baited one of two pots placed to the left and right. These were head-turn/gaze cues, not an isolated test of eye movement with a motionless head.

The experiment combined sustained (called dynamic) versus momentary gaze with ostensive versus non-ostensive attention-getting signals. Ostensive here means signals addressing the receiver as a communication partner. In the publisher’s introduction, sustained gaze remains directed at the target until choice; momentary gaze briefly indicates the target before returning to the cat. The university account describes 24 trials, six per combination. Full details of cue timing, counterbalancing, food-odor controls and exclusions cannot be audited from the available preview and are not assumed here.

Findings — Lumi

• The publisher abstract reports about 70% overall success at group level; its Results preview gives 70.42% and reports performance significantly above chance. The preview says 18 cats met its individual above-chance criterion across the 24 trials. Group-level success does not mean every cat demonstrated the same reliability.
• The authors report no significant effect of gaze type or ostensive addressing on choice success. This is a failure to detect a difference under these conditions, not proof that the conditions are equivalent.
• Ostensive addressing shortened the latency to establish eye contact (abstract: p = 0.006). Latency means time until an event begins. This is an attention measure, distinct from choosing the indicated pot.
• The abstract describes only a trend toward higher success in young cats (p = 0.085). It should not be presented as an established age advantage.

Lumi’s interpretation: The results support useful cross-species cue following in this tested group. They do not by themselves establish that cats understood a person’s intention to inform them, read a mind, or possess a particular theory of mind. The university account itself raises learned directional associations as an alternative to understanding intent.

Limitations and inference boundaries — Lumi

The primary limitation of this synthesis is partial source access. The inspected excerpts do not let us fully assess alternative cues, detailed trial order, recruitment losses, coding or statistical models. An unavailable control is not necessarily an absent control: those are questions for the full Methods, not accusations about the study.

The requested food motivation and home-companion sample constrain generalization to other cats and settings. Repeated trials and an aggregate success rate should not be translated into a universal species-level ability. The gaze cue included head movement, so “cats read eyes alone” overstates what was tested.

The authors discuss domestication and socialization as possible explanations. This study did not experimentally isolate evolutionary selection from individual learning history. Likewise, faster eye contact does not demonstrate a deeper understanding of communicative intention, and a nonsignificant choice contrast does not prove ostensive cues never matter.

Reading route — Lumi

Start with the publisher Abstract → Introduction’s definitions of gaze types → available Success snippet → university explanation of the two-pot setup. Keep two columns: time to look at the person and correct pot choice. If full text becomes available, read complete Methods and controls before extending the inference. The expanded route below stays within accessible evidence.

Guided reading — Lumi

Original explanatory synthesis by Lumi, based on the publisher preview and authors’ university account. This is not a full-text audit or original scientific research.

1. Identify what actually moves

• What to look for: The university description of an experimenter’s head turn toward one of two pots, and the introduction’s sustained versus momentary gaze definitions.
• What it means: The cat receives a visible directional cue involving the head and gaze together. Momentary means the cue ends before the eventual choice; sustained means the direction remains available until choice.
• Why it matters: “Following human gaze” is an operational description of this cue, not proof that the animal can extract eye direction independently of the head. Isolating those components would require a different comparison.

2. Separate attention from information use

• What to look for: The abstract reports faster eye contact with ostensive addressing but no detected improvement in choice success.
• What it means: An attention-getter can change when a cat starts attending without changing which location it ultimately selects. Latency and accuracy answer different questions.
• Why it matters: Collapsing them into one “communication score” would hide the most informative contrast. The result is not simply “speaking to cats works” or “speaking makes no difference”; it depends on the outcome being measured.

3. Read the group result without turning it into every cat

• What to look for: About 70% success in the abstract and 18 individually above-chance cats in the accessible Success snippet. Also notice the recruitment request for food motivation.
• What it means: Average performance can support a group-level inference while individual certainty differs. An individual not crossing a statistical threshold has not thereby been proven incapable.
• Why it matters: Neither “all cats can reliably do this” nor “the other cats cannot follow gaze” follows from these numbers. Selection, motivation and repeated observations need to stay visible in the interpretation.

4. Mark the boundary between an observation and its explanation

• What to look for: The university discussion contrasts understanding intent with learning that an indicated direction is worth following. The publisher offers domestication and socialization as broader possibilities.
• What it means: Successful choices are observed behaviour; the internal mechanism and its evolutionary origin are further explanations requiring additional comparisons. Full procedural controls remain outside the inspected preview.
• Why it matters: A strong reading ends with a specific, bounded claim: this group used the provided directional cues successfully. Ask what control would separate competing explanations, rather than inserting an unverified control or a human-like mental state into the paper.

Two reflection questions — Lumi

1. How can an attention-getting signal make eye contact faster without improving the final choice? What separate measures would you keep?
2. What comparison would help distinguish following a head direction from understanding that a person intends to provide information?