You were taught that vergence is a motor reflex. A target appears at a new distance, the eyes register disparity, the disparity drives a motor command, the eyes converge or diverge to close the gap. Stimulus in, movement out — a closed loop with the retina as the trigger. That model treats vergence the way you'd treat a knee-jerk: input arrives, muscle responds, nothing upstream gets consulted. It's clean, but it's framed incorrectly. Vergence isn't waiting for disparity to tell it what to do. It's a cortically generated behavior that disparity merely refines.
Kumar, Han, Garbutt, and Leigh's 2002 study is the cleanest demonstration. Ten adults alternated fixation between a near and a far target on a predictable 1.25-second cycle. Vergence began before the target actually jumped in 83% of divergence trials and 70% of convergence trials — median lead, 191 milliseconds. Make the timing unpredictable and the anticipatory responses nearly disappear. There was no disparity signal yet when the eyes started moving. What triggered the movement was an expectation, built from a learned temporal pattern, not a retinal event. That's not a reflex responding early. That's a plan being executed on schedule.
Where does that plan come from? Gamlin and Yoon found a region of frontal cortex, immediately anterior to the classic saccadic frontal eye field, containing neurons that fire specifically for vergence and accommodation — a cortical site dedicated to forming a depth-and-focus goal, distinct from the tissue that plans conjugate gaze shifts. This is the part of the standard teaching that gets skipped entirely: vergence has its own frontal representation. It isn't a downstream consequence of the oculomotor system, and it isn't a downstream consequence of the retina's feedback information either. It's planned.
This is also why vergence responds to things that have nothing to do with binocular disparity. Sperandio and colleagues showed that vergence angle shifts when attention is captured — including by an auditory cue that carries no retinal disparity information at all. Attention alone, directed toward or away from a location, changed the state of the vergence system. Tzur and colleagues found the same attentional modulation was measurably weaker in children with ADHD, which only makes sense if vergence is downstream of an attentional/cortical gating mechanism in the first place. If vergence were purely disparity-driven, an auditory attention cue should do nothing to it. It does something to it.
Voluntary control makes the same point more directly. People can converge their eyes on instruction, with no target present and no disparity to respond to — Khan and colleagues used exactly this paradigm in an fMRI study, having subjects converge on an auditory "go" cue after the visual target had already disappeared. The vergence system executed the command anyway, and it produced its own distinct cortical activation pattern, separate from what disparity-driven convergence produces. This is vergence behaving like a motor act you can initiate at will, the way you initiate a reach — not like a reflex you can only trigger by presenting the right stimulus.
Then there's proximal vergence, which the field has known about since Maddox but rarely built into how it teaches the reflex model beyond just saying how it is something that can affect phorometric data. Schor and colleagues showed that simply changing a subject's perceived distance to a target — through size, blur, and context, independent of any real disparity change — was enough to drive a substantial vergence response, on the order of several prism diopters per meter of perceived distance. The eyes converged because the person believed the object was closer. Belief, not disparity, was doing the driving.
Put this together and the architecture stops looking like a reflex arc and starts looking like what it is: a cortically initiated behavior shaped by what the patient is attending to, what they've been instructed to do, what they expect to happen next, and what they believe about the scene in front of them — with disparity acting as an error signal that fine-tunes an already-initiated plan rather than as the thing that starts the movement. Frontal cortex forms the goal. Attention gates and modulates it. Instructional set can override it entirely. Perceived distance alone can generate it. Disparity comes in afterward to correct it.
This matters clinically because it changes what a "vergence problem" can actually be. A patient can have a completely intact disparity-vergence reflex — normal amplitude to a controlled prism step in the exam chair — and still struggle badly with vergence in real tasks, because real tasks aren't controlled prism steps. Reading requires constant depth-and-direction planning under sustained attentional load. A patient whose attentional system is taxed, distracted, or dysregulated may show a vergence system that works fine in isolation and falls apart the moment it has to be voluntarily initiated, sustained, and coordinated with what they're supposed to be doing. That's not a fusional weakness. That's a cortical planning and attentional-gating problem wearing a vergence-symptom costume.
The instinct in the exam room is to test vergence as a motor output and stop there, because that's the model we were handed — present the stimulus, measure the reflex, done. But if vergence is cortically generated, then how you instruct the patient, what you ask them to attend to, and what they expect from the task are not neutral details around the measurement. They're part of what's being measured. The eyes aren't just responding to what's in front of them. They're doing what the person's mind has already decided to do.