The Colour of the Letter A

ASynaesthesia — the condition in which a letter reliably arrives with a colour, or a Tuesday with a position in space — poses an awkward problem for a science built on public evidence: the only witness to the experience is the person having it. The instrument that made the field respectable is disarmingly simple. A participant is asked to choose the colour of each letter, sent away, and asked again months later without warning; people with the condition reproduce their choices with 85 per cent agreement or better, while volunteers who were told to invent associations and memorise them rarely exceed 40 per cent. On the strength of that test, prevalence in the general population is now usually put at around 4 per cent. What the test cannot say is where any association came from, and it is over that question, rather than over whether the experiences are real, that two research programmes have spent a decade talking past each other.

BLars Salminen, a neuroscientist at the Fenwick Laboratory for Sensory Neuroscience, went looking for the machinery. Between 2015 and 2022 her group scanned 96 grapheme-colour synaesthetes while they read plain black letters on a white screen, and found activity in V4, the patch of cortex that handles colour in everyone, at a level those same letters never produced in matched controls. The step that made the work matter came next. When brief magnetic pulses were delivered to V4, damping it for roughly 400 milliseconds, the participants' colours faded and their scores on a task that pits letter against ink dropped to the level of people with no synaesthesia at all. Salminen is exact about the reach of this. The same interference left sequence-space forms, in which the months of the year occupy fixed positions in the air, entirely intact, and those, her group later showed, run through parietal rather than visual cortex. The demonstration is of one variant, not of synaesthesia in general.

CThat account carries a prediction, and Salminen has tested it. If a variant depends on two regions of cortex lying close enough to leak into one another, the pairings people report should cluster where the anatomy permits and be scarce where it does not. Her group catalogued 84 published variants and scored each for anatomical plausibility: 71 per cent linked regions that are neighbours, while forms requiring a signal to cross the brain — a taste that arrives with a touch on the left hand — accounted for 3 per cent of reports and never once survived a consistency test in her own laboratory. She presents the pattern as a constraint on one explanation rather than as a census of the condition, and adds that a form may be rare for reasons owing nothing to wiring: nobody, she observes, has yet counted how often the stranger pairings simply go unmentioned by the people who have them.

DRavi Sundaram, a developmental psychologist at the Corbett Institute, thinks the scanner has been aimed at the wrong end of the process. Cross-activation, he argues, describes the state of an adult brain; it says nothing about how a particular letter came by a particular colour, and the answer to that, in a startling number of cases, is a toy. In 2019 his group collected the letter-colour sets of 217 adult synaesthetes born in the 1970s and 1980s and found that 24 of them matched, letter for letter, a plastic alphabet sold in enormous quantities during those decades; the odds of that arising by chance he puts at below one in a million. A follow-up study tracked 1,600 children through primary school and reported that the pairings of those who eventually met the consistency criterion had drifted for two or three years before hardening. He concedes two weaknesses. The toy match explains a minority even of his own sample, and children who already see colours may simply play with such sets more.

EThe dispute has not stayed inside the laboratory. A commercial memory course launched in 2021 advertised that its exercises could give anyone a synaesthete's recall, citing a training study in which 47 adults who drilled letter-colour pairs for five weeks reproduced them at 76 per cent some months later — short of the diagnostic threshold, and, on the study's own later follow-up, gone within a year in most participants. Salminen points out, awkwardly for her own side, that nothing in her stimulation experiments shows whether trained associations run through the same cortex as lifelong ones, because a mechanism established in one group is not a licence to assume it in another. Sundaram replies that the two questions cannot be held apart once money is involved, since a company will read any neural finding as an endorsement. Both object to the same habit above all: the appetite for one spectacular case in place of a count of what ordinary participants do.

FWhat is emerging is not a winner but a division of labour. Salminen's programme explains how a private colour becomes a public physical event, and explains equally why that explanation stops at the edge of a single route through the brain; Sundaram's establishes where the content of an association comes from, and his answer is: from the world outside, far more often than anyone expected, though not in every case. Neither result makes the other unnecessary. Neither settles the question a school or a clinic actually faces, which is whether an association strong enough to show up in a scanner is ever useful enough to be worth cultivating in a child who does not have one. No study now running is designed to answer that. What both reject is the assumption that made the subject famous, namely that such an experience must be either an accident of biology or a trick of memory.