What makes you one person?

Flash a word to the left visual field, which reaches only the right hemisphere, and the patient says they saw nothing, because speech is usually a left-hemisphere function. Ask the left hand, controlled by the right hemisphere, to point at what was shown, and it points correctly.

Findings like this, accumulated over decades, produced the standard account: the operation leaves two systems that perceive, decide and act separately, and only one of them talks.

The account is in the textbooks. The data behind it, the team that tested it again in 2017 pointed out, were never reported as numbers.

Outside the laboratory these patients do not behave like two people sharing a body. They report feeling unchanged since the operation, and they say they see the whole visual field and control the whole body. The exceptions reported are anecdotes from the first period after surgery, when a hand may act outside the speaking person’s control.

What nobody has done is watch their ordinary lives systematically, so this rests on what they say and on how they seem in company.

Any account has to explain both the laboratory results and the ordinary life, and the popular version of this story only explains the first.

In 2017 a group tested two split-brain patients systematically, varying which visual field a stimulus appeared in and which hand or voice answered. If each hemisphere were a separate conscious agent, responses should have been accurate only when the responding side matched the side that saw.

They were not. Both patients detected and located stimuli anywhere in the visual field, with either hand and by speaking. The patient tested further also identified stimuli that way, while still showing the classic inability to compare two of them across the midline. The authors described it as divided perception with undivided consciousness.

The detection task used both patients. The four experiments after it, on comparison, identity, orientation and location, used one of them.

That is a direct test of the standard account rather than a reinterpretation of old data, and it is two patients out of a population that is small and shrinking. The reading is contested. Volz and Gazzaniga argue that hemispheres separated for many years have had time to learn to cue each other, through small movements of the eyes or face, so answers that arrive unified need not come from a unified subject. The 2017 team had tried to rule that out. Schechter and Bayne argue that even unified answers would show at most one agent, not one stream of experience.

If unity does not depend on that cable, something else supplies it. Three candidates are argued over: information crossing below the cortex, one hemisphere driving the hand on its own side, and two separated halves that have learned to signal to each other. None of them is settled.

Gazzaniga’s answer is that the self was always constructed rather than located. That is not a dodge. In his account a left-hemisphere interpreter builds a story about events that goes beyond the information it has, and Nisbett and Wilson argued in 1977 that people asked why they did something often offer a plausible theory rather than a report of the process.

There is no agreed method for counting subjects of experience in a brain, intact or divided. The dispute above is between careful researchers looking at the same kind of patient, which is what it looks like when a question has no agreed measurement rather than no data. Researchers from both sides of it wrote a joint review in 2020 and agreed that the evidence is not yet enough to answer it.

Until there is such a method, the number of minds in a split brain is a question about evidence for behaviour and a question about experience, and the two are not the same question.