Birds / Interface study

Giving parrots a choice to connect

Source year 2024 · Note reviewed 2026-10-14

In a six-month, nine-parrot home study, Table 2 lists 65 parrot-triggered live calls and 40 playback calls out of 108 possible in each phase. Recalculation from the displayed counts gives live-minus-playback differences of +50, +17, +75, +17, +8, -8, -8, +33 and +25 percentage points for P1–P9. The printed row is reversed for seven birds and appears to swap the P3/P5 entries. The authors report a live-phase advantage, but the published duration totals do not reconcile across the text and Table 3; the duration magnitude remains unresolved.

What we reviewed · Methods reviewed

Published paper pages 3–12 reviewed for participant screening/ethics, protocol, analysis, results and limitations. Page 9 was rendered and Table 2's P1–P9 differences recalculated from the displayed session counts. The current University of Glasgow deposit record and Crossref DOI relation metadata were checked. No raw-data reanalysis or replication.

Not yet independently reviewed by an avian, animal-computer-interaction or statistical specialist.

Study at a glance

Study unit: Nine home-kept parrots over six months.
Interface: The parrot rang a bell and selected a bird image; a caregiver then placed the call.
Design: Six live sessions came before six playback sessions. Sessions allowed up to two five-minute calls in a three-hour window; condition order was not counterbalanced.
Measures: Bird-triggered call counts and duration, time to first call, caregiver diaries, interviews and questionnaires.
Welfare support: Caregivers received welfare training; specialists reviewed recordings; birds could retreat behind a visual barrier and caregivers were instructed to stop calls if distress appeared.

Reading location: CHI 2024 published version: participant/method sections and analysis, pages 3–8; Tables 2–4, page 9; results and limitations, pages 10–12.

What this does — and doesn’t — establish

The live phase always came first, so condition order was not counterbalanced; reciprocal calls, caregiver cues and variable home devices also complicate interpretation. Table 2 labels the comparison ‘Live minus playback,’ but the printed values correspond to playback-minus-live for P1, P2, P4 and P6–P9; P3 and P5 also appear transposed relative to the displayed counts. No underlying data were available to verify author intent, and no correction was found in the current University of Glasgow record or Crossref relation metadata; that check is not exhaustive. The Data Analysis text reports 561 live-phase and 142 playback minutes, while summing Table 3's displayed per-parrot cells gives about 239 and 149 minutes, respectively; no underlying data were available to resolve this. The inferential tests were not independently recomputed. Engagement is not a direct welfare, comprehension or language measure.

What we can build on

The parrots signaled and selected a caller, but caregivers initiated the calls. Treat animal choice, caregiver mediation and responsiveness as separate design variables; preserve retreat and stopping options. This is an animal-computer-interface study, not a passive acoustic-collection or communication-decoding result.

The next reading task

Obtain independent avian animal-computer-interaction and statistical review, and author clarification if contact is separately authorized, before reusing participant-level Table 2 values or adapting the protocol. Keep the printed results distinct from our arithmetic check; do not silently correct the paper.

Go to the sources

Parrot video-call primary paper ↗
Hirskyj-Douglas, Cunha & Kleinberger (2024). Call of the Wild Web: Parrot Engagement in Live vs. Pre-recorded Video Calls. CHI 2024, Article 625, 1–14. DOI: 10.1145/3613904.3641938. The University of Glasgow record lists CC BY-NC-ND 4.0 and a Data Availability Statement of ‘No’.
Published full text used for the methods review ↗

Original editorial note by Talk2Nature contributors. AI-assisted, source-grounded synthesis; not a peer review or an independent replication. Linked works retain their own rights.