The Sensory Connotation of Meaning and the Cues That Fit a Brand

Abstract

Brands use names, designs, sounds, and scents to express abstract meanings. This article defines sensory connotation as a meaning's position on attributes such as dark–bright, angular–round, and warm–cool, and uses language models to estimate it at scale. The method uses a compact set of fixed axes that can score new words, phrases, and cue libraries without a new human norming study for every target. We test each claim against evidence that the scoring models did not produce. Human raters placed 80 concepts on the same seven axes and agreed with the instrument at r = .80. A single model predicted held-out human means more closely than an individual rater (mean r = .77 vs. .51), and a four-model consensus reached r = .79. Profile similarity predicted how similar concept pairs felt beyond embeddings, valence, arousal, dominance, and human free association (b = .146, p < .001). Across 6,236 commercial perfumes, names, disclosed notes, and bottle designs showed corresponding profiles across language and pixels. Randomly assigned profile-matched scent descriptions received higher fit ratings than mismatched descriptions. In a preregistered study, profile fit predicted larger stated store-scenting allocations beyond expected pleasantness. Delivered-odor choices showed a similar, though less conclusive, pattern, including a significant preregistered supporting within-name test. A label-blind "blind painter" tests what the profile itself contains by rendering cues after the target label is removed. Together, these results show how sensory connotation can measure meaning, test whether that measure captures shared structure, and rank candidate cues before those cues are built.