Five Is the Wrong Number

Aristotle listed five senses and we have been repeating him for two and a half thousand years. The list is wrong, and not by a little.

Proprioception tells you where your limbs are without looking, and is the reason you can touch your nose with your eyes shut. The vestibular system reports balance and acceleration. Interoception reports hunger, thirst, heartbeat, breath and the state of your organs, and is increasingly thought to underpin a great deal of emotional experience. Nociception handles pain, which is not a form of touch and runs on different fibres. Thermoception handles temperature, separately from pressure.

Of these, proprioception is arguably the one whose absence is most disabling. There are a small number of documented cases of people who lost it through nerve damage, and they must watch their limbs continuously to control them; looking away means the arm is simply gone. Walking has to be rebuilt as a deliberate visual task.

We do not notice these senses because they never fail, which is a bad reason not to count them.

Now colour, which is the case where the constructed nature of perception is easiest to demonstrate.

Surfaces reflect a continuous spectrum of wavelengths. The human retina samples that spectrum with three cone types whose response curves overlap heavily, and everything we experience as colour is computed from the ratios between three numbers.

Two consequences follow immediately. First, entirely different physical spectra can produce identical triples and therefore look the same (metamers) which is why a fabric can match under shop lighting and clash in daylight, and why a camera and an eye can disagree about a colour. Second, colour is not a property sitting in the object waiting to be detected. It is what this particular three-sample system constructs from the physics. A mantis shrimp with twelve receptor types is sampling the same world; whatever it produces is not a more accurate version of what you see.

The dress, in 2015, was a natural experiment in exactly this.

The photograph was badly lit and genuinely ambiguous about the illumination. You could not tell from the image whether the dress was in cool daylight or under warm artificial light. Your visual system has to assume something in order to discount the illuminant and report the surface, and it normally does this so smoothly that you never notice the assumption is there.

Here, viewers made different assumptions. Those whose systems assumed cool bluish daylight discounted blue and saw white and gold. Those assuming warm indoor light discounted yellow and saw blue and black. Two populations, same photons, different colours, and both entirely sincere. The argument was so ferocious precisely because nobody could locate the step where the disagreement happened; it occurred below anything introspection can reach.

Here is one more that is easy to miss: the present moment is assembled.

Light and sound travel at different speeds and are processed at different rates. Touch from your toe takes longer to arrive at the brain than touch from your cheek, because the nerve is longer. Yet a clap you see and hear is experienced as one event, and touching your nose and your toe simultaneously feels simultaneous.

The system appears to buffer inputs across roughly a tenth of a second and align them before delivering a unified moment. Which means that what you experience as "now" is a slightly retrospective edit of inputs that have already happened: a construction rather than a live feed, assembled for coherence rather than for timeliness.

And the sense that the body is yours is also computed, which the rubber hand illusion shows with unusual clarity.

Hide a participant's real hand, place a rubber hand in front of them in a plausible position, and stroke both with brushes in synchrony. Within a minute or two most people report that the rubber hand feels like theirs. Threaten it with a needle and they flinch, with a measurable skin conductance response. Ask them to point at their real hand without looking and they point toward the rubber one.

Body ownership is not a given fact that perception reports. It is an inference from the correlation between what is seen and what is felt, continuously recomputed, and it can be extended to an object that is obviously not you as soon as the correlations line up.

Which leaves a small, useful piece of humility for arguments about taste. Variants near the OR6A2 olfactory receptor gene are associated with detecting aldehydes that coriander shares with some soaps, and the frequency of those variants differs sharply between populations. When someone tells you coriander tastes of soap, they are not being difficult and they are not making a judgement you could talk them out of. They are receiving a different signal from the same leaf.