Pierre Gloor's book The Temporal Lobe and Limbic System has an excellent discussion of this pages 316-318. I'll try to type it in by hand and stick it on paste

Pierre Gloor's book The Temporal Lobe and Limbic System has an excellent discussion of this pages 316-318. I'll try to type it in by hand and stick it on pastebin later since it is unfortunately a rare out of print book (Google Books has it but it seems to show much less text these days than it used to). Gloor suggest that because low level olfaction is not able to distinguish parts of a smell it is more heavily associated with context than other senses. He notes a study where recognition of olfactory cues are not as immediately reliable as visual or auditory cues but there is very little decay over time of that recognition ability in olfaction while visual and auditory cues are recognized no better than chance after 3 months. He compares the Proust description to electrical stimulation of the amygdala in an epilepsy patient and suggests that particularly strong memories may be amygdala assisted recall (and notes that similarly vivid recall can happen with other senses but seems to be less common).