Grapheme-color synesthesia is an atypical condition in which individuals experience sensations of color when reading printed graphemes such as letters and digits. For some grapheme-color synesthetes, seeing a printed grapheme triggers a sensation of color, but hearing the name of a grapheme does not. This dissociation allowed us to compare the precision with which synesthetes are able to match their color experiences triggered by visible graphemes, with the precision of their matches for recalled colors based on the same graphemes spoken aloud. In six synesthetes, color matching for printed graphemes was equally variable relative to recalled experiences. In a control experiment, synesthetes and age-matched controls either matched the color of a circular patch while it was visible on a screen, or they judged its color from memory after it had disappeared. Both synesthetes and controls were more variable when matching from memory, and the variance of synesthetes' recalled color judgments matched that associated with their synesthetic judgments for visible graphemes in the first experiment. Results suggest that synesthetic experiences of color triggered by achromatic graphemes are analogous to recollections of color.

Grapheme-color synaesthesia is an atypical condition characterized by coloured sensations when reading achromatic text. Different forms have been characterized, but this is somewhat controversial. In associative grapheme-colour synaesthesia, written graphemes can automatically trigger a sensation of colour in the ‘mind's eye’, but hearing the name of a grapheme does not. This allowed us explore the precision with which synaesthetes match triggered synaesthetic colours across separate presentations, versus the precision for recalled experiences cued by spoken graphemes. We recorded CIE coordinates, and found that matches for triggered sensations were equally variable relative to recalled experiences. To ensure this was not due to insensitivity of our apparatus, we next had synaesthetes and age-matched controls either match the colour of a circular patch while they could see it, or from memory after it had disappeared. Both synaesthetes and controls were more variable when matching from memory, and synaesthetes were more precise when matching colour hue, but not brightness. Interestingly, the variance of synaesthetes' recalled matches in this experiment matched that associated with synaesthetic colours in the first experiment. Overall, our data suggests that, for associative grapheme-colour synaesthetes, synaesthetic colours behave more like recalled colours, as opposed to physical colours that can be seen.