Neural mechanisms linking odour, emotion and memory: a systematic narrative review of 119 studies with quantitative synthesis and certainty-of-evidence appraisal

Background. Olfaction is the only sense with access to the amygdala and entorhinal cortex after just two synapses, without passing through the thalamus. This anatomical feature is often used to explain why odours are so tightly bound to emotion and memory, but the evidence is scattered across molecular biology, systems neuroscience, memory psychology and applied research of widely varying quality.

Objectives. (1) To describe the neural mechanisms linking odour to emotion (amygdala, OFC) and to memory (entorhinal–hippocampus); (2) to quantify the magnitude and certainty of behavioural and clinical effects; (3) to evaluate popular applied claims about fragrance.

Methods. A narrative systematic review with descriptive quantitative synthesis (following the SWiM guideline). 119 papers (39 from the original list, 80 from targeted searches, with DOIs verified) were read in detail; 95 papers were available in full text (75 complete, 20 partially incomplete) and 24 as abstract only. Data were extracted using a 25-field schema; level of evidence was rated with OCEBM; certainty of evidence per outcome was assessed according to GRADE principles. No new meta-analysis was performed because of heterogeneity in outcomes and designs.

Results. fMRI/MVPA shows that the posterior piriform cortex encodes perceptual quality rather than chemical structure, whereas single-cell recordings show that the piriform still decodes chemical identity better and that it is the hippocampus that leans towards perceptual identity; of 5 regions recorded at single-cell level, the amygdala is the only one that encodes valence (r = 0.56), and it encodes intensity only when the odour has valence; the OFC represents positive and negative value simultaneously and is modulated by labels/state; the hippocampus prioritises the first odour–object association and predicts memory one week later. Targeted memory reactivation (TMR) during sleep has a small effect (g = 0.29; 95% CI 0.21–0.38), falling to 0.13 after correction for publication bias; odour cueing gives g = 0.32 (0.15–0.50), no different from sound. Odour-evoked autobiographical memories are more emotional (d = 0.47) and less often rehearsed (d = 0.52) but are evoked less successfully than with words/pictures (63% vs 76%). Chemosignals of fear have a small-to-moderate effect (g = 0.36; 0.31–0.41). In applications, inhaling lavender essential oil reduces situational anxiety (g = −0.64 in RCTs not at high risk of bias), but 89% of RCTs are at high risk of bias and the effect is not significant in mild anxiety; one RCT with good blinding control found lavender no different from water for mood, cortisol or pain. Olfactory training improves TDI (g = 1.10), improving identification/discrimination more than threshold (g = 0.34) — suggesting (inference) a large role for central learning. Impaired odour identification predicts Alzheimer’s dementia (lowest quartile HR = 10.3). Ambient scent in retail increases recall (r = 0.18) but does not significantly increase spending (r = 0.05).

Conclusions. The odour–emotion–memory link has a solid neural basis, but it does not simply come from a “shortcut into the limbic system”: odour perception is itself a memory process, carries valence from the olfactory bulb onwards, and is strongly governed by expectation and context. Outcomes with moderate certainty are TMR, fear chemosignals, the predictive value of odour identification for Alzheimer’s disease, oral Silexan and recall in retail; olfactory training and the mood-lifting effect of lemon have low-certainty evidence; the therapeutic claims of aromatherapy are mostly of low or very low certainty.