Particle.news

Dogs Show Humanlike Consonant Bias When Segmenting Speech

The result suggests simple pattern learning can produce a consonant-driven word‑segmentation bias in dogs, leaving open whether domestication or an older mammalian hearing mechanism explains it.

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.