If you’ve ever taken a psych class, you’ve met the memory “two-drawer” model. One drawer is episodic memory: the stuff of lived experience—your tenth birthday party, the first day at a new job, the moment you realized you’d left your phone in the cab. The other is semantic memory: facts and general knowledge—Paris is in France, bees make honey, and yes, you recognize that swoosh as Nike.
Neat, tidy… and possibly a little misleading.
A new brain-imaging study in Nature Human Behaviour reports that when people successfully retrieve an episodic memory versus a semantic one, the brain may not flip between two separate systems the way many researchers have long assumed. Instead, the same broad set of brain areas appears to do the heavy lifting in both cases.
The test: facts vs experiences, matched as closely as possible
One reason this debate keeps resurfacing is that it’s hard to compare episodic and semantic memory fairly. Studies often use very different tasks—different materials, different difficulty, different instructions—then interpret any differences as “episodic vs semantic.”
So the team—based at the University of Nottingham and the University of Cambridge—designed the two tasks to be as similar as they could make them. They recruited 40 participants and used a simple, surprisingly relatable stimulus: logos and brand names. In the semantic task, people saw a logo and had to recall the associated brand from their existing knowledge of the world. In the episodic task, people learned logo–brand pairings during an initial study phase, then later had to recall those pairings from that recent learning episode.
While participants tried to remember, the researchers measured brain activity using fMRI (functional magnetic resonance imaging), which tracks changes in blood flow that correlate with neural activity.
The result: the “memory boundary” got blurry
Here’s the headline: when memory retrieval was successful, the researchers found no clear difference in neural activity between the semantic and episodic tasks. They report that the brain networks and activation patterns most often linked to retrieval didn’t reliably separate into “fact mode” versus “event mode.”
The lead author, psychologist Roni Tibon, put it bluntly in the accompanying press release: “We were very surprised… a long-standing research tradition suggested there would be differences in brain activity… but… we found that the distinction didn’t exist.”
That doesn’t mean episodic and semantic memory are identical in every way—patients with certain brain injuries can show uneven damage to one kind of memory over the other, and memory scientists have good reasons for the classic split. But it does suggest that the act of pulling something up—the search, the reconstruction, the “Is this right?” feeling—may rely on shared machinery more than we’ve been comfortable admitting.
WHAT IS A REGISTERED REPORT?
A Registered Report is a publishing format designed to reduce cherry-picking and “pretty result” bias. Instead of journals judging a study mainly by whether the findings are exciting, researchers submit their research question, methods, and analysis plan up front—before they collect data (or before they analyze it). Peer reviewers then critique the design, not the outcome. If the plan is strong, the journal gives in-principle acceptance, meaning it will publish the paper regardless of whether the results are positive, negative, or null—so long as the authors follow the approved plan. That makes surprising “no difference” findings more trustworthy.
A better metaphor: one search engine, different queries
Think of memory retrieval less like opening two different filing cabinets, and more like using one search engine.
Sometimes you type a query like “my tenth birthday”—and the result is a scene, a timeline, a place, a feeling. Sometimes you type “that logo with the bitten fruit”—and the result is a fact: Apple.
The new study’s claim is that the search engine itself may be largely the same, even if the databases it’s scanning—and the kind of output you get back—feel wildly different from the inside.
Or, to put it another way: the difference might lie more in what you’re retrieving than in which retrieval machine you’re using.
Why this matters (and what it doesn’t prove)
If the brain doesn’t cleanly separate episodic and semantic retrieval, that changes what a “good” memory study looks like. Instead of treating the two as different planets with different physics, researchers may need more within-study designs that test them side-by-side, under closely matched conditions.
The team also points to long-term implications for understanding memory disorders. Tibon notes that these findings could help researchers rethink how memory is affected in conditions like dementia and Alzheimer’s—though it’s important to underline that this study doesn’t test a treatment or diagnose disease. It’s a basic science map of what retrieval looks like in healthy volunteers.
There are also real limits to remember. First, it’s one kind of semantic knowledge. Logos and brands are a specific slice of “facts.” Semantic memory also includes word meanings, concepts, and categories—things that might behave differently. Also, fMRI has blind spots. It’s powerful, but it’s not a direct readout of neurons firing, and subtle differences can be hard to detect. And finally, “No difference” is a careful claim. The researchers used statistical tools aimed at weighing evidence for “no meaningful difference,” not just shrugging at a non-significant result—but scientific confidence still grows through replication and varied tasks.
Still, the core takeaway is satisfyingly provocative: your brain may retrieve the name behind a logo using much of the same architecture it uses to retrieve a personal moment. Different kinds of remembering, more shared scaffolding than we thought.
Endnotes
- University of Nottingham press release, “New study redefines our understanding of how memory works,” posted on EurekAlert! (January 27, 2026).
- Tibon, R., Greve, A., Humphreys, G., et al. “Neural activations and representations during episodic versus semantic memory retrieval.” Nature Human Behaviour (published January 27, 2026). DOI: 10.1038/s41562-025-02390-4.