Overview
- The peer‑reviewed study published in Science on Sept. 24, 2026 recorded noninvasive EEG from 20 adult humans and 20 companion dogs while they listened to continuous streams of three-syllable nonsense words.
- Neural measures — including intertrial coherence (neural entrainment) and an event‑related potential around 400 milliseconds — showed stronger word‑level tracking for consonant-structured streams in both species.
- Dogs tracked consonant-defined word patterns even when vowels were made louder and more acoustically prominent, ruling out a simple loudness explanation for the bias.
- The consonant bias appeared in dogs regardless of early-life exposure to human speech, with animals raised in shelters or on the street showing similar neural signatures to those raised in speech-rich homes.
- Humans still outperformed dogs on detecting vowel-based patterns and on synchronizing to finer speech-sound rhythms, and the authors say follow-up work is needed to test whether the bias arose through domestication or reflects an ancient mammalian auditory ability.