AAN 2026: Why Alzheimer Disease Brain Stimulation Trials Have Had Mixed Results—And How a New Memory Network Could Fix That

In an interview with Pharmacy Times at the 2026 American Academy of Neurology (AAN) Annual Meeting, Calvin W. Howard, MD, a clinical scientist from Brigham & Women's Hospital and instructor of neurology at Harvard Medical School, explained that convergent causal mapping combines evidence from stroke lesions, deep brain stimulation (DBS), and transcranial magnetic stimulation (TMS) to overcome the anatomical blind spots of each individual modality, ultimately revealing a distributed brain network for memory that extends well beyond the hippocampus to include parietal, lateral temporal, and even cerebellar regions.

Howard described how applying this convergent memory network to 19 TMS studies and 2 DBS studies in Alzheimer disease revealed that trials whose stimulation sites were more strongly connected to the network tended to achieve better outcomes, offering a mechanistic explanation for the historically inconsistent results in the field.

Pharmacy Times: What is convergent causal mapping, and why was combining lesion, DBS, and TMS data essential to finding it?

Calvin W. Howard, MD: Convergent causal mapping is a technique we can use to take evidence from things that cause a symptom to get worse or better and to figure out where they live in the brain. Think of strokes that cause what I'm going to talk about—memory loss. They can occur, of course, in the hippocampus and the mesial temporal lobe, but they can occur all over the rest of the brain as well. However, those strokes tend to fall only within areas that blood vessels supply, so we can only really probe the anatomy related to blood vessels with stroke-related memory loss. What about things like transcranial magnetic stimulation—a high-intensity magnetic pulse outside of the head that generates an electrical current inside the head? That can alter activity as well, and it allows us to test regions that don't typically stroke, like the parietal lobe, which is often in a well-supplied watershed territory. The same is true with deep brain stimulation. It allows us to reach very deep nuclei or white matter structures—like the fornix—and evaluate how those are related to memory. So each of these modalities causes a symptom or causes that symptom to improve, and we can combine them to cover the blind spots of each individual approach. DBS can accommodate for the fact that strokes only cover vascular territories, and together they allow us to gain a more holistic understanding of where things live in the brain.

Pharmacy Times: Your network extends well beyond the hippocampus—what does that mean for how we've traditionally thought about memory localization in the brain?

Howard: Memory localization in the brain has been a somewhat evolving concept over the last 100 years or so. At the most fundamental level, we know the hippocampus is important—if it is removed, infected, or severely damaged, people develop very severe memory dysfunction. However, we know through clinical experience that patients with strokes, stimulation, or other interventions across the brain can develop memory deficits to varying degrees. Moving through the localization of human memory, we have the hippocampus, the mesial temporal lobe, the fornix, and the circuit of Papez, and then we actually have distributed regions beyond that which act in concert to support human memory—such as parietal attention networks, executive control, and the ability to focus and summon something to memory; aspects of the lateral temporal cortex, which are very important to memory; and even components of the cerebellum, as we found in our paper. Memory at its core is fundamentally a way of being able to capture something and then summon it to mind at will. But there are many different regions across the brain that each have their own part to play in the overall process of learning something and then recalling it later. This distributed brain network gives us a better understanding of the regions that are consistently involved in that overall process.

Pharmacy Times: How does this convergent network explain the inconsistent outcomes seen in prior Alzheimer disease stimulation trials?

Howard: Rather surprisingly, after we found this convergent network across approximately 1,300 patients with variable memory outcomes following lesions or stimulations, we went and applied it to Alzheimer's disease brain stimulation trials—19 TMS studies and 2 DBS studies. The outcomes had been quite mixed throughout these prior publications, with some being very positive and others being less compelling. When we mapped the location of stimulation sites across all of these studies and related them to the convergent memory network, we found that the more a given study happened to be targeting regions strongly connected to the convergent memory network, the better its overall outcomes were. It is not a perfect relationship, but it helps us understand a meaningful degree of the variance in those outcomes.

Pharmacy Times: How close are we to translating this into optimized brain stimulation protocols for memory dysfunction?

Howard: At the moment, we have the convergent memory network as a theoretical target. There are regions on the surface of the brain that are strongly connected to it, and there are regions deep within the brain that are strongly connected as well. On the surface, the parietal lobe is a very accessible place to stimulate with TMS, and there is a peak within the parietal lobe that is very strongly connected to the convergent memory network. I am working with Joe Taylor at Harvard Medical School to develop a prospective trial—at least a feasibility study for now—evaluating what happens when we administer very high-dose transcranial magnetic stimulation targeting the convergent memory network directly in patients with Alzheimer disease. We have found that this type of approach is very effective for treating symptoms like depression and anxiety, and network-guided targeting appears to improve TMS outcomes. We are hopeful this could help with Alzheimer disease as a noninvasive treatment option. On the invasive side, there are primarily deep brain stimulation approaches for Alzheimer disease. Two groups have handled most of the invasive stimulation trials, and we are going to work with some of their patients in concert with collaborators in Cologne who have already undergone the procedure and have DBS electrodes sitting within the fornix. Previously, we had no clear framework for programming those patients—the electrodes were implanted, turned on at a reasonable estimated current, and the outcomes were generally underwhelming. The hypothesis is that if we bring these patients back, rearrange their stimulation parameters, and move them from a suboptimal location on the electrode to one that more maximally engages the convergent memory network, we may be able to acutely improve their memory performance. Finally, I am working directly with a neurosurgeon colleague at Harvard Medical School, and we are investigating—in the context of patients who have stereotactic EEG electrodes placed throughout the brain—whether those who happen to have electrodes near the convergent memory network peak, which lies in the posterior mesial temporal lobe just behind the hippocampus, show improved memory when those electrodes are activated. If so, this could represent a potentially efficacious target for future therapeutic approaches.