At least some insects can retain memories formed during their caterpillar stage into adulthood, even after the dramatic brain remodeling that occurs inside a chrysalis. The best evidence comes from a study on tobacco hornworm moths, not butterflies specifically, but the finding upended the long-held assumption that metamorphosis wipes the neural slate clean. The reality, though, is more complicated than a simple yes or no. Whether a memory survives depends on the species, the type of memory, and even the age of the caterpillar when it learned.
The Experiment That Changed the Conversation
For most of the twentieth century, scientists assumed that metamorphosis was too destructive for any larval memory to survive. The brain of a caterpillar is substantially dismantled and rebuilt during the pupal stage, so the idea that learned information could persist through that process seemed implausible. Then, in 2008, researchers at Georgetown University published a study that offered striking evidence to the contrary.
The team worked with Manduca sexta, the tobacco hornworm, a large caterpillar that metamorphoses into a hawk moth. They exposed late-stage caterpillars to a specific odor paired with a mild electric shock, training them to associate that smell with something unpleasant. The caterpillars learned to avoid the odor, which was no surprise. The surprise came later: after the caterpillars pupated and emerged as adult moths, they still avoided the same odor they had been trained to dislike as larvae.1PubMed Central. Retention of memory through metamorphosis: can a moth remember what it learned as a caterpillar?
The researchers were careful to rule out a simple chemical explanation. If residual odor molecules had stuck to the pupa’s surface from the training environment, the moths might be reacting to a lingering smell rather than a genuine memory. To test this, they applied the training odor directly to the pupae of untrained caterpillars. Those moths showed no aversion. They also washed the pupae of trained caterpillars and found the aversion persisted anyway. The memory was in the brain, not on the body.1PubMed Central. Retention of memory through metamorphosis: can a moth remember what it learned as a caterpillar?
Why This Finding Was So Surprising
To appreciate why the Manduca result was remarkable, you need a rough picture of what metamorphosis does to an insect’s nervous system. The process is not a gentle renovation. During the transition from larva to pupa, some neurons die outright. Others are stripped of their existing connections: the branching structures that receive and transmit signals are pruned back to stubs and then regrown in entirely new configurations suited to the adult body. On top of that, entirely new neurons that played no role in the larva’s life mature and integrate into the adult brain.2PubMed. Developmental neuroethology of insect metamorphosis
Work on the cabbage white butterfly (Pieris brassicae) showed something similar at the cellular level: motor neurons and association neurons in the thoracic nerve centers break down during metamorphosis, while small precursor cells divide and give rise to new neural tissue.3Journal of Insect Physiology. Changes occurring in the central nervous system of Pieris brassicae L. (Lepidoptera) during metamorphosis The adult brain is not just a redecorated version of the larval brain; it is substantially rebuilt.
So how could a memory survive all that demolition? The key seems to lie in the neurons that are conserved and remodeled rather than destroyed. Not every larval neuron dies during metamorphosis. Some are retained, pruned back, and then regrown with new adult-appropriate wiring. If a memory is stored in one of those conserved neurons, or in a network of them, it could theoretically ride through the pupal stage intact even as the connections around it are reconfigured. The Manduca study provided the first direct behavioral evidence that this actually happens.
When the Caterpillar Learns Matters
One of the more interesting details in the Manduca study was that the age of the caterpillar at the time of training made a difference. The researchers trained caterpillars at the fifth instar, which is the final larval stage before pupation. Caterpillars trained at this late stage retained their odor aversion as adult moths. The implication is that the brain structures encoding the memory at this late developmental point are among those that survive metamorphosis.
Earlier training might not produce the same result. During the earlier larval stages, the nervous system is still growing and changing substantially, and the neurons involved in a memory formed at that point may be more likely to be among those dismantled during metamorphosis. This lines up with what we know about which neurons are conserved: the ones that persist tend to be more mature, well-established cells rather than newly formed or rapidly developing ones.2PubMed. Developmental neuroethology of insect metamorphosis A caterpillar that learns something in its last few days of larval life may have the best chance of carrying that knowledge into adulthood, because the neural circuits encoding the memory are more likely to be among the survivors.
This suggests that the phenomenon is not some universal property of insect metamorphosis but rather a consequence of specific developmental timing. The door through which a memory can pass is narrow and depends on which neurons happen to store the information and whether those neurons make it through the pupal overhaul.
Not Every Insect Keeps Its Memories
The Manduca result, while compelling, does not appear to be universal across insects. A more recent study in Drosophila, the common fruit fly, found no evidence that aversive memories survive metamorphosis. Researchers trained fly larvae to associate a particular smell with an unpleasant stimulus, the same basic paradigm used in the moth experiments. When those larvae emerged as adult flies, the learned aversion was gone.4PubMed Central. No evidence for aversive associative olfactory memory through metamorphosis in Drosophila
The Drosophila team went a step further. A brain structure called the mushroom body is central to learning and memory in insects. During metamorphosis, the mushroom body neurons from the larval stage are pruned back and regrown, much like other neurons. The researchers used genetic tools to largely prevent this pruning and regrowth, essentially trying to keep the larval mushroom body wiring intact through the pupal stage. Even with this intervention, no larval memories could be detected in the adult flies.4PubMed Central. No evidence for aversive associative olfactory memory through metamorphosis in Drosophila
This is a significant counterpoint. It tells us that the physical preservation of neurons is not, by itself, enough to guarantee memory survival. Something about the way Manduca stores or encodes memories may differ from how Drosophila does it. Or perhaps the difference lies in the degree of nervous system remodeling: flies and moths diverged hundreds of millions of years ago, and their metamorphic processes, while superficially similar, involve different lineages of neurons and different patterns of conservation and destruction. The honest summary of the science is that memory retention through metamorphosis has been demonstrated in at least one moth species but has failed to replicate in fruit flies, and we do not yet know where most butterfly and moth species fall on that spectrum.
What About Butterflies Specifically?
Here is where the evidence gets thin. The Manduca study used a moth, and the Drosophila study used a fly. Butterflies and moths are closely related, both belonging to the order Lepidoptera, and they undergo very similar metamorphic processes. It is reasonable to suspect that whatever mechanism allows Manduca moths to retain memories may also operate in some butterflies, but that has not been directly demonstrated with the same experimental rigor.
The cellular-level work on Pieris brassicae, a true butterfly, confirmed that its nervous system undergoes the same kind of breakdown and rebuilding seen in moths.3Journal of Insect Physiology. Changes occurring in the central nervous system of Pieris brassicae L. (Lepidoptera) during metamorphosis That means the same obstacle to memory retention exists: neurons being pruned, rewired, and in some cases destroyed. But it also means the same potential escape route exists, via neurons that are conserved and remodeled rather than killed. Whether specific butterfly species actually retain larval memories through that escape route remains an open experimental question.
Part of the reason the data are sparse is practical. Manduca sexta is a large, hardy, lab-friendly insect that has been used in neuroscience research for decades. Many butterfly species are harder to rear in controlled conditions, harder to train in behavioral assays, and more fragile. Running the kind of odor-aversion experiment that worked beautifully with Manduca is considerably more difficult with a painted lady or a monarch. The absence of evidence for butterflies is not evidence of absence; it reflects the difficulty of the experiments more than the biology.
Do Food Preferences Carry Over?
A related question people often ask is whether a caterpillar’s diet shapes the adult insect’s behavior. If you grew up eating milkweed, do you prefer to lay your eggs on milkweed? This idea has a name in entomology: the Hopkins’ host selection principle, which proposes that female insects prefer to lay eggs on the same type of plant they fed on as larvae. If it held up broadly, it would be a form of larval experience shaping adult behavior, which is conceptually related to memory retention.
The evidence, however, mostly goes the other way. A study on the comma butterfly (Polygonia c-album) found that what the caterpillars ate had no effect on where the adult females chose to lay their eggs.5Ecological Entomology. No effect of larval experience on adult host preferences in Polygonia c‐album (Lepidoptera: Nymphalidae): on the persistence of Hopkins’ host selection principle Similarly, a study on the fall webworm moth found that females overwhelmingly did not distinguish between host plants and non-host plants when choosing where to lay eggs, and largely did not prefer their natal host.6PubMed. Plant oviposition preferences in a generalist herbivore: tests of host plant selection and the Hopkins’ host selection principle
There is one interesting wrinkle from the fall webworm work. The species has two morphotypes (red-headed and black-headed), and among the small number of females that did lay eggs on a plant rather than on a non-plant surface, the red-headed morphotype did show a preference for its natal host. But the overall pattern was stark: most females ignored plant identity entirely.6PubMed. Plant oviposition preferences in a generalist herbivore: tests of host plant selection and the Hopkins’ host selection principle The researchers suggested that the lack of selectivity may actually be what allows generalist species to exploit many different host plants rather than being locked into one.
So while a specific odor aversion can apparently make it through metamorphosis in at least some moths, the broader question of whether larval feeding experience shapes adult egg-laying behavior does not have strong support. These are different phenomena. An odor aversion is a learned response to a negative stimulus, while host-plant preference involves a complex set of sensory cues, instinctive drives, and ecological pressures. The fact that one can survive metamorphosis does not mean the other routinely does.
What Kind of “Memory” Are We Talking About?
It is worth being honest about what “memory” means in this context, because it is not the same as human autobiographical memory. Nobody is claiming that a moth emerges from its cocoon and thinks, “I remember that time I got shocked as a caterpillar.” The type of memory demonstrated in the Manduca experiments is associative: the caterpillar learned to link a specific sensory cue (an odor) with a negative outcome (a shock), and that association persisted. This is closer to the way your hand flinches away from a hot stove than to the way you recall your childhood.
Insects do not have the neural architecture for episodic memory, the rich, contextualized recall that lets you replay past events in your mind. What they do have are efficient circuits for associating stimuli with outcomes, and these circuits are surprisingly sophisticated. The mushroom bodies, the brain structures most associated with insect learning, can encode multiple overlapping associations and modify behavior based on experience. The question of whether memories survive metamorphosis is really a question about whether the physical substrate of these associations, likely patterns of synaptic strength in specific circuits, can persist through a period of massive neural renovation.
For Manduca, the answer appears to be yes, at least for associations formed late in larval development. For Drosophila, the answer appears to be no, even when researchers tried to physically preserve the relevant circuits.4PubMed Central. No evidence for aversive associative olfactory memory through metamorphosis in Drosophila That discrepancy hints that the mechanism of memory survival, if it exists, is more subtle than simply keeping the right neurons alive. It may involve molecular changes within neurons, changes to gene expression, or storage in parts of the nervous system that have not yet been identified as memory-relevant.
Why the Question Captivates Us
There is something philosophically striking about the idea that an animal can go through a transformation as radical as metamorphosis and emerge with any thread of its former self intact. The caterpillar’s body is broken down into a kind of cellular soup inside the chrysalis; organs dissolve and reform. The fact that the nervous system is not entirely rebuilt from scratch, that some neurons are conserved and carry forward into the adult, provides a biological bridge between two radically different life stages. Whether or not a given butterfly species demonstrably retains larval memories, the physical continuity of at least part of the nervous system means that the adult is not a wholly new organism. It is the same individual, remodeled.2PubMed. Developmental neuroethology of insect metamorphosis
The popular version of this story, the one that circulates on social media, often overstates the evidence. Headlines like “butterflies remember their caterpillar days” gloss over the fact that the key study was done in moths, that it involved a very specific type of learned aversion, that timing mattered, and that the result has not replicated in fruit flies. The real picture is more interesting precisely because it is incomplete. Scientists have cracked a window into how insect brains handle the extraordinary challenge of metamorphosis, and what they have found so far is that the answer varies by species, by memory type, and by developmental stage. The general public tends to want a clean yes or no, but the science here is still being mapped out, one species and one experiment at a time.
Implications for How We Think About Insect Minds
Research on memory through metamorphosis connects to a larger, ongoing shift in how scientists think about insect cognition. For a long time, insects were treated as essentially robotic, running fixed behavioral programs with little capacity for individual learning or flexibility. That view has been crumbling for decades, and the Manduca memory result is one more piece of evidence against it. If a moth can carry a learned association through the most radical bodily transformation in the animal kingdom, the neural machinery underlying insect learning is more robust and more deeply embedded than the old “simple automaton” model would predict.
This does not mean insects have rich inner lives or anything approaching human consciousness. But it does mean their nervous systems are doing something more interesting than running pre-written scripts. Bees learn the locations of flowers and communicate them to hive-mates. Ants adjust foraging strategies based on past success. And at least one moth species carries learned associations through a period when much of its brain is disassembled and rebuilt. The boundaries of what counts as “memory” and “learning” in animals without backbones continue to expand as researchers develop better tools to ask the question.