Contribution
It shows in a moth that an associative memory formed in the caterpillar is still expressed in the adult. It rules out the standard rival, a “chemical legacy” of odour carried on the pupa, with two manipulations of the pupal case. And it shows that what survives depends on when it was learned: an aversion learned at third instar lasts through two larval molts and is lost in metamorphosis, while one learned at fifth instar is kept.
Key insight
Metamorphosis does not erase the animal’s past wholesale, and it does not keep it wholesale either. It keeps what was learned late. The authors read this through the mushroom bodies: in Drosophila the earliest lobe (γ) is pruned in pupation, and a later one (α′/β′) keeps its projections. So their result “challenge[s] a broadly-held popular view of lepidopteran metamorphosis: that the caterpillar is essentially broken down entirely, and its components reorganized into a butterfly or moth” (p. 5).
Assumptions
- Position in a Y maze after ten minutes measures preference. Each animal was scored, blind, as in the odour arm, the ambient-air arm or “no choice” (p. 6). The paper does not say how “no choice” animals enter the proportions.
- Shock is an ecologically fair aversive stimulus. The authors argue this from mantids learning to avoid milkweed bugs and hornworms sensitized by pinching (p. 5).
- Chemical legacy means residue on or in the pupa. Washing tests the outside of the pupal case. The inside is addressed by design: a gaseous, ephemeral odour and a shock, not an ingested chemical (p. 4).
- Manduca mushroom bodies develop as Drosophila’s do, scaled to five instars instead of three. The mechanism section is explicitly conditional on this (“If M. sexta MB development is analogous”, p. 5).
Key results
- Training (pp. 5–6). Two to three days after the molt to fifth instar, up to ten larvae at a time on conductive agarose: 10 s of ethyl acetate (EA), then 10 s of odour with a 16–18 V AC shock. Eight bouts, an hour apart, in one day. Larvae were tested the next day, adults three days after eclosion. At least three cohorts per treatment; sample sizes set by power analysis.
- Larvae learn (p. 2). Fifth instars avoided EA only after forward pairing: 78%, N = 41, p < 0.001 (two-tailed binomial against 50%). Untreated (N = 46, p = 0.32), shock only (N = 43, p = 1.0), odour only (N = 29, p = 1.0) and backward pairing (N = 22, p = 0.52) showed nothing.
- The memory survives larval molts (p. 2). Larvae trained in early third instar avoided EA 10 to 14 days later as late fifth instars: 81%, N = 32, p < 0.001. Not different from fifth-instar training (χ² = 0.1128, df = 1, p = 0.74).
- It survives metamorphosis (pp. 2–4). Adults tested 28 to 35 days after larval conditioning. Controls chose at random (N = 31, 23, 28; each p = 1.0). Forward-paired adults avoided EA: 77%, N = 27, p = 0.005. The authors note 78% of larvae and 77% of adults: “a long-lasting and stable memory” (p. 5).
- Individuals keep their choice (pp. 2–4). Of forward-paired adults that chose ambient air, 80% had chosen it as larvae (N = 15, p = 0.035). Control groups’ constancy did not differ from 50%.
- Early learning does not survive metamorphosis (p. 4). Adults trained as third instars did not avoid EA (N = 15, p = 0.5).
- Not chemical legacy (p. 4). EA gel applied to naive pupae, four to six times from three weeks after pupation, gave no aversion (N = 30, p = 0.20). Washing the pupal cases of trained larvae, one and three weeks after pupation, did not remove it (N = 31, p = 0.438). Washing control pupae changed nothing (N = 25, p = 0.423).
- The proposed carrier (p. 5). In Drosophila the γ lobe forms in the embryo and is pruned in pupation; α′/β′ forms in mid-third instar and keeps intact projections; α/β begins at pupation. With five instars, Manduca third-instar training likely precedes α′/β′ neurogenesis. So early memory would sit in γ and be lost, late memory in α′/β′ and be kept. The results are “consistent with, but do not provide conclusive support for” surviving synaptic connections (p. 5). The test they propose is to ablate α′/β′.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Manduca larvae form an associative odour aversion from forward pairing of odour and shock | strong within the design: four controls, including backward pairing | p. 2, Fig. 2 |
| C2 | An aversion learned in the fifth instar is expressed by the adult moth | moderate: one group of 27, one odour, one lab | pp. 2–4, Fig. 3 |
| C3 | The adult aversion is not chemical legacy | moderate: external residue tested both ways; internal residue argued from design, not tested | p. 4, Fig. 4 |
| C4 | Individuals, not just groups, keep their larval choice | weak to moderate: 12 of 15 | pp. 2–4 |
| C5 | An aversion learned at third instar survives larval molts but not metamorphosis | moderate for the molts (N = 32); weak for the loss, a null in 15 adults with no direct comparison to the fifth-instar adults | pp. 2, 4 |
| C6 | The memory is carried by late-forming mushroom-body neurons whose projections survive pupation | hypothesis: by analogy with Drosophila; the Manduca brain was not examined | p. 5 |
Concepts
- holometabolous: undergoing complete metamorphosis, larva to pupa to adult.
- chemical legacy (Corbet 1985): an adult’s response shaped by odours from the larval environment met at eclosion, not by memory.
- Hopkins’ host-selection principle: the idea that adults prefer the host they fed on as larvae. The authors call the evidence for it “equivocal” (p. 4).
- constancy: the proportion of individuals choosing ambient air as larvae that also did so as adults.
- forward and backward pairing: odour before shock, which conditions, and shock before odour, which here did not.
- mushroom bodies (γ, α′/β′, α/β lobes): paired brain structures receiving antennal-lobe input, the presumed seat of olfactory memory.
Connections
- Alloway 1972 (LIT-1781), unread. Cited (ref. 36) among a century of studies of larval experience and adult behaviour in beetles, moths and butterflies. The paper claims to be the first conclusive demonstration in Lepidoptera, not the first in insects. It counts Drosophila (Tully et al. 1994) and a weevil among controlled earlier cases (p. 4).
- Blackiston, Shomrat and Levin 2015 (LIT-1707) restate the result in one sentence and give a different account of α′/β′ remodelling, as “partial pruning” rather than intact projections. Blackiston moved to Levin’s lab.
- Shomrat and Levin 2013 (LIT-1732), the planarian counterpart. There the claim is that memory lives outside the brain; here the evidence points to neurons that survive inside it.
- Dias and Ressler (NOTE-1281) also used an odour paired with shock. They carried a sensitivity across generations, through sperm. This paper carries a learned aversion across one life’s metamorphosis, and does not test offspring.
Bearing on the record
- QUESTION-302, what carries it across. The paper excludes one
answer and points to another. These are inferences beyond the authors’
claims.
- Residue on the pupa is excluded. Washing and scenting were each tested the other way round, and neither moved behaviour.
- The timing result is the most informative fact for the question. A memory that survives two larval molts and then fails in metamorphosis is not stored in “the animal” as such. It is stored in something that the molts keep and metamorphosis replaces. The Drosophila mushroom-body map supplies a candidate, a lobe that survives, but this paper only fits the map; it does not test it.
- So on this evidence the carrier is a surviving part of the nervous system, not a body-wide store. That is the opposite of the planarian reading the question also holds.
- QUESTION-302, whether the moth is the one that learned. The
moth is the same organism as the caterpillar on any biological view.
The paper adds a thin psychological connection: one learned aversion,
held by about three quarters of a group and kept by 12 of 15 tracked
individuals.
- That fits THEORY-464’s line that the relation that matters comes apart from numerical identity. Here identity holds and connectedness is partial and depends on when the learning happened. A third-instar caterpillar is connected to its moth less than a fifth-instar one, though both are the same animal.
- It does not make the Parfitian case (LIT-469) of full psychological continuity through a wholly rebuilt body. The body is not wholly rebuilt; the carrier, if the authors are right, is the part that is not.
- The offspring claim in the question, from the unpublished Kobe repetition, is not tested here at all.
- The owner’s thesis (CLAIM-491). No direct bearing. No dated note is proposed for it.
- A dated note is proposed for QUESTION-302 (section 3).
- No THEORY filed. One study; the claim waits on the mechanism.
- No ML instruction.
Limitations
- One species, one odour, one lab, one aversive stimulus. Adult groups are 15 to 31.
- The “no choice” animals are scored but their handling in the proportions is not stated. The figure captions give “means±SD” for what are proportions of binary choices, without saying over what.
- Internal residue is argued away, not tested. The authors fault an earlier study for possible residue “in the insect haemolymph” (p. 4). Their own answer is the gaseous stimulus, not a measurement.
- The washing test’s statistic is unclear. The text says washing did not cause “a loss of odor aversion in adults (N = 31, p = 0.438)” (p. 4). If that p were the binomial against 50% used everywhere else, it would mean the washed adults showed no significant aversion. The Figure 4 caption marks some values as differing from random. So the 0.438 is presumably a comparison with unwashed adults, but the test is not named.
- The loss after third-instar training is a non-significant result in 15 adults. The two adult groups are not compared directly.
- The mechanism is borrowed. No Manduca brain was examined.
- Inconsistencies and typos.
- Figure 3’s caption (p. 3) gives shock only N = 23 and odour only N = 28; the text (p. 2) gives odour alone N = 23 and shock alone N = 28.
- Figure 4’s caption (p. 4) gives “N = 40 adults, 30 larvae” and “N = 39 adults, 31 larvae”, while the text gives N = 30 and N = 31 for the adults. The caption’s labels look swapped.
- “fifth insta” (Fig. 3 caption, p. 3); “holometablous” and “implications. as retention” (p. 5); “thoratic” and “resulted in in a change” (p. 6).
Open questions
- Does ablating the α′/β′ lobe, or its Manduca equivalent, abolish the adult memory of fifth-instar training and spare nothing else?
- Is the haemolymph free of the odour? A direct measurement would close the last chemical route.
- Does an appetitive memory, or one for a second odour, survive the same way? One aversion is a narrow sample of what a caterpillar learns.
- Does the aversion reach the moths’ offspring? The Kobe repetition claims so for swallowtails; it is unpublished.