A dog’s head direction was tracked by AI during the Bouba/Kiki experiment at the University of Parma.
*A University of Parma study shows dogs look longer at rounded shapes when hearing “Bouba” and at spiky shapes when hearing “Kiki”. The experiment isolates canine sound‑shape association from owner influence.*
A new study suggests dogs share a human‑like instinct for matching sounds to visual forms. Researchers at the University of Parma played the nonsense words “Bouba” and “Kiki” while displaying rounded and spiky shapes on opposite screens. The dogs’ gaze patterns were captured by a GoPro and later analysed with AI to pinpoint where they were looking.
Seventy domestic dogs participated across two trials. In the first, 51 dogs sat with their owners while the sounds and shapes alternated. In the second, 19 dogs performed the same task but their owners wore headphones playing rock music, eliminating any auditory cue from the humans. The design isolates whether the dogs respond to the sounds themselves or to subtle owner signals.
Animal behaviorist Eleonora Biffi recruited 70 domestic dogs and placed each pair with its owner in a room containing two screens and speakers. Over roughly 15 minutes, the speakers alternated the nonsense words “Bouba” and “Kiki” while the screens swapped a rounded shape and a spiky shape between left and right. A GoPro recorded each dog’s head movements. Researchers fed the footage into an AI that tracked head orientation, using it as a proxy for gaze direction. The AI output allowed statistical testing for patterns that rise above random chance.
The analysis revealed a clear bias: dogs looked significantly more at the rounded shape while “Bouba” played, and they looked at and explored the spiky shape when “Kiki” was played. This pattern emerged across the 51 dogs in the first experiment, confirming a sound‑shape correspondence that mirrors the well‑documented human effect. The result suggests dogs may process auditory‑visual pairings in a way comparable to humans, despite the words being meaningless.
To test owner influence, the second experiment removed auditory cues from humans. Nineteen dogs performed the same task while their owners wore headphones blasting rock music, preventing them from hearing “Bouba” or “Kiki”. In this condition the general bias toward spiky shapes vanished, indicating that the earlier spiky‑shape preference likely stemmed from subtle owner cues rather than an innate canine bias. The authors argue that because humans typically favor rounded shapes, owner influence is an unlikely driver of the round‑shape preference.
The findings open a window onto why dogs are quick to pick up human language cues. If dogs naturally align sounds with visual properties, they possess a cross‑modal mapping ability that could have facilitated early domestication. The study stops short of claiming definitive equivalence with humans, but it provides a measurable foothold for future work on canine cognition, auditory processing, and the evolutionary roots of interspecies communication.
Yes. The study found dogs looked significantly more at the rounded shape when 'Bouba' was played and at the spiky shape when 'Kiki' was played.
Seventy domestic dogs in total: 51 in the first experiment and 19 in the second.
In the first experiment owners sat with their dogs and could hear the sounds; in the second they wore headphones playing rock music to block the sounds and remove any potential cue.
The study stops short of declaring a universal canine sound‑shape lexicon, but it stakes a claim that dogs, like humans, can link arbitrary phonetics to visual form. As AI‑driven tracking refines behavioral readouts, future experiments may map the neural pathways behind this ability. Until then, the Bouba‑Kiki effect adds a new layer to our understanding of why dogs are such attuned companions.
Sources: ScienceAlert article by Jess Cockerill, Proceedings of the Royal Society B: Biological Sciences (Biffi et al., 2026).