IQ Metrics
IIF Certified Assessment Start IQ Test
IIF Certified Assessment
Data studyBrain and cognition · Version 2026.11

Can plants and slime moulds learn? Every experiment, and whether it was repeated

We found every published experiment that claims learning, memory or habituation in an organism with no neurons, 1906-2026, and scored each first report on whether an independent laboratory repeated it.

About this study

Report type
Data study: our ledger of published experiments on learning without neurons, scored
Data
50 experiments (30 first reports, 20 tests and follow-ups), 1906-2026; plants, slime moulds, amoebae, ciliates, fungi
Sources
Europe PMC searches (6,205 records), 137 abstracts and texts read, PubMed Central full texts, OpenAlex citations
Years
First reports 1906 to 2026; searches and citation counts as of 10 November 2026
Licence
CC BY 4.0 (our dataset and charts)
Version
2026.11
Sources
69
Short answer

According to the authors, some can: of 30 first reports of learning, memory or choice in organisms with no neurons, 5 were independently repeated in the same species, and 4 of those repeats found the effect. The one repeat on a plant, the pea, found none. IQ Metrics Research ledger, 1906-2026, as of 10 November 2026.

30 first reports5 of 30 independently repeated4 of 5 repeats found the effectCC BY 4.0
5 of 30first reports independently repeated in the same species
4 of 5of those repeats found the effect
1 of 10plant claims independently repeated (pea; not found)

01 Key findings

  1. Few results have been checked by someone else. 5 of 30 first reports were repeated by an independent laboratory in the same species; 4 of those repeats found the effect.IQ Metrics ledger, 1906-2026
  2. The ciliates fare best. Stentor roeseli (1906), Stentor coeruleus (1988) and Paramecium caudatum (2006) were each repeated by an independent laboratory, with the effect found.Dexter, Rajan, Alipour
  3. Plants have one independent test and it found no effect. The 2016 pea associative-learning claim was not reproduced in 2020; the original authors dispute the replication’s protocol and the replicator replied.Markel 2020; Gagliano et al. 2020
  4. Slime moulds: one same-species repeat. Physarum habituation to bitter bridges was reproduced by a second laboratory; the maze, the trail and the decoy results were not (the decoy effect was tested in a different slime mould).Yoneoka 2023; Yin 2025
  5. Most tests are new. 10 of the 11 independent tests were published in 2020 or later; the median wait for a first repeat is 16 years.IQ Metrics calculation
  6. Attention does not follow checking. The most cited claim (the 2000 maze study, 1,032 citations) has no independent repeat in our ledger.OpenAlex

02 Statistics to quote

5 of 30

first reports of learning, memory or choice without neurons were repeated by an independent laboratory in the same species; 4 of the 5 repeats found the effect.

IQ Metrics ledger
22 of 30

first reports have no independent test of any kind in our searches: no same-species repeat and no other-species test.

IQ Metrics ledger
30

first reports in plants, slime moulds, amoebae, ciliates and fungi, 1906-2026; 50 experiments counting the tests that followed.

IQ Metrics ledger
1 of 10

plant claims was repeated by an independent laboratory: pea associative learning, and the repeat found no effect.

IQ Metrics ledger
3 of 4

ciliate claims (Stentor and Paramecium) were repeated by an independent laboratory, and all three repeats found the effect.

IQ Metrics ledger
16 years

median wait between a first report and its first independent repeat (range 4 to 113 years, n = 5).

IQ Metrics calculation
113 years

between Jennings’ 1906 report of a hierarchy of responses in Stentor roeseli and the 2019 repeat with the correct species.

Dexter et al. 2019
10 of 11

independent tests were published in 2020 or later; the first one in our ledger is from 2019.

IQ Metrics ledger
300 micrometres

long: Paramecium caudatum, the smallest organism in the ledger whose claimed learning was repeated by an independent laboratory with the effect found.

Alipour et al. 2022
0 vs 302

neurons: every organism in the ledger has none; the nematode Caenorhabditis elegans, the reference animal, has 302 and habituates to taps.

White et al. 1986; Rankin et al. 1990
1,032

citations of the most cited claim, the 2000 maze-solving study (Nakagaki et al.); it has no independent repeat in our ledger (OpenAlex, 10 November 2026).

OpenAlex
25 of 30

first reports record behaviour (movement, growth direction, choice); 5 record gene activity or electric charge, called learning or memory by their authors.

IQ Metrics ledger

03 Can plants and slime moulds learn?

According to the authors of the experiments, some organisms with no neurons can: we found 30 first reports of habituation, associative learning, memory or choice in plants, slime moulds, amoebae, ciliates and fungi, published 1906-2026, and only 5 of them were repeated by an independent laboratory in the same species. Four of those five repeats found the effect and one, on pea seedlings, did not. The ledger takes no side in the “plant neurobiology” dispute and makes no claim about what any plant or cell experiences: it counts what was published, who tried to repeat it and what the repeat found.

We built our own ledger because the nearest existing work is review articles, not a count. Gershman and colleagues revisit single-cell learning, centred on the 1950s Paramecium experiments of Beatrice Gelber [60]; Reid reviews cognition in the slime mould Physarum polycephalum [64]; Adelman reviews classical conditioning in plants and writes that the reported results are “mixed, with no replications of positive results” [61]. None of them lists every experiment with the same written rule for “was it repeated”. We searched eight ways, read 137 records and kept 50 experiments, scoring each first report by a rule we wrote before computing any verdict.

Organism with no neurons
A plant, fungus, slime mould or single-celled animal. It has no nerve cells and no brain; the nematode Caenorhabditis elegans, with 302 neurons, is the contrast here [69].
Habituation
A response that weakens when a harmless stimulus is repeated and comes back after a pause, not just because of fatigue. It is the simplest form of learning and the one most often repeated here [29] [45].
Associative learning
A cue that was neutral starts to trigger a response because it was paired with another stimulus, as in Pavlov’s dogs. Claims of it without neurons are the most disputed [60].
Memory (as used here)
Information from an earlier event is kept and changes a later response: a count of touches in seconds [10], a trace left in the environment [27], or habituation that survives a month of dormancy [32].
Slime mould (Physarum polycephalum)
A single giant cell with many nuclei that spreads over a surface as a network of tubes. It is the model organism of most non-neural learning claims [64].
Ciliate
A single-celled animal covered in beating hairs, such as Stentor (a trumpet-shaped cell 1-2 mm long when extended) and Paramecium [46] [50].
Independent repeat
A later paper that runs the same design on the same species and states whether the effect was found, written by authors who are not the first or last author of the original. It is our test of “someone else tried it”.
Plant neurobiology
A name for research on plant signalling and behaviour. Some authors argue plants are intelligent [66]; others argue they neither possess nor require consciousness [67]. This ledger counts experiments and takes no side.

04 How many experiments claim learning in organisms with no neurons?

We found 30 first reports published from 1906 to 2026, and 20 later experiments that repeated, tested in another species, followed up or extended them: 50 experiments in all. They come from 6,205 records in eight Europe PMC searches, of which 951 passed a written word rule and 629 survived scripted blockers; we read 137 by hand (abstract or full text) and kept 50 rows. The count is a lower bound: a paper our searches missed is not in it, and we list every record we read and dropped.

Figure 1

From 6,205 records to 50 experiments: how we built the ledger

Stage 1: unique records from eight Europe PMC searches (plants, slime moulds, ciliates and amoebae, fungi, bacteria, "without neurons" wording). Stage 2: a written word rule on title and abstract. Stage 3: scripted blockers (immune cells, animals with brains, disease, machine learning). Then 137 records picked by hand and read; 50 became rows, of which 30 are first reports. As of 10 November 2026.

Source: Europe PMC [1]; PubMed [2]; screening by IQ Metrics (data/screen_decisions.tsv)

Data table
stagecountpct_of_firstpct_of_previous
Records from 8 Europe PMC searches6205100100
Pass our scripted word rule95115.315.3
Left after scripted blockers62910.166.1
Picked by hand for an abstract or text read (with leads)1372.221.8
Rows in the ledger500.836.5
Claims (first reports)300.560
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

Eight of the 50 rows were not found by the searches themselves but by named leads and the citation lists of the originals: that is how Dexter and colleagues’ 2019 confirmation of Jennings’ 1906 Stentor result [42] and the Armus laboratory’s later negative result [49] came in. The older literature is the biggest gap: 25 records from 1956 to 1982 on habituation and learning in Spirostomum, Stentor, Vorticella and Paramecium have no abstract in PubMed and no open text, so we know of them only from titles and a review [60] and did not count them.

Figure 2

The 49 experiments since 1980: first reports and the tests that came after

Every row of the ledger from 1980 on, by publication year and organism group; colour = first report, independent test (effect found or not found) or same-laboratory follow-up. 10 of the 11 independent tests are from 2020 or later. The one row before 1980, the 1906 Stentor report of Jennings, is in the trail figure further down. As of 10 November 2026.

Source: Our ledger of 50 experiments [1] [2]

Data table
grouptypefromtolabeldetail
Plantsevent2008Volkov 2008Volkov 2008 (Dionaea muscipula): The trap sums small electrical charges that alone do not close it and shuts when a threshold is reached, which the authors call short-term electrical memory.
Plantsevent2009Volkov 2009Volkov 2009 tests Volkov 2008: found. Measured the charge that closes the trap: 8 microcoulombs (small trap), 9 (large trap), 4.1 at 28-36 degrees C; the cumulative effect points to short-term electrical memory.
Plantsevent2014Gagliano 2014Gagliano 2014 (Mimosa pudica): Leaf-folding in response to repeated drops habituates, more strongly and for longer in costly environments, and the plants still showed the learned response after a month undisturbed.
Plantsevent2016Gagliano 2016Gagliano 2016 (Pisum sativum): Pea seedlings trained in a Y-maze with a fan (neutral cue) before light grew toward the fan alone, which the authors call associative learning.
Plantsevent2016Dener 2016Dener 2016 (Pisum sativum): Split-root pea plants put more root growth on the variable side when mean nutrients were low and on the constant side when high, as risk-sensitivity theory predicts; the authors note non-adaptive mechanisms may explain part.
Plantsevent2016Böhm 2016Böhm 2016 (Dionaea muscipula): The trap counts touch-induced action potentials: the jasmonate pathway switches on after the second stimulus and digestive genes after more than three; a single touch is "memorized" as a subthreshold signal.
Plantsevent2019Mittelbach 2019Mittelbach 2019 (Nasa poissoniana): Stamens presented pollen earlier after flowers had been touched at fixed intervals; the authors report first evidence of pre-emptive stamen presentation based on remembered visit intervals.
Plantsevent2020Markel 2020Markel 2020 tests Gagliano 2016: not found. A replication with a larger sample and blinded scoring found no significant difference between trained and control plants.
Plantsevent2020Bhandawat 2020Bhandawat 2020 (Arabidopsis thaliana): Seedlings conditioned with music and heat switched on heat-response genes more strongly than with music alone; the authors call this associative learning (gene expression, not movement).
Plantsevent2023Iosip 2023Iosip 2023 tests Böhm 2016: found. A Venus flytrap mutant still fires touch action potentials but cannot count them and does not enter the hunting cycle; the authors describe a calcium clock for memory and decision.
Plantsevent2024Ponkshe 2024Ponkshe 2024 tests Gagliano 2016: not stated. Reports 11 practical issues met while trying to replicate the pea experiment; the abstract gives no effect result.
Plantsevent2024Salvalaio 2024Salvalaio 2024 (Arabidopsis thaliana): Roots exposed for hours to a weak electric field overshoot and then habituate in their bending, and past exposures change the later response (hysteresis).
Plantsevent2025Farkas 2025Farkas 2025 (Solanum lycopersicum): Tomato plants conditioned with ultrasound and drought used one stimulus as a predictor of the other in 3 of 4 target genes; transcript levels also showed habituation or sustained induction (gene expression, not movement).
Plantsevent2025Vishton 2025Vishton 2025 (Mimosa pudica): Over 15 days of two lit days then one dark day, leaf movements shifted to follow the pattern; the authors say movement tracks the number of illumination events.
Slime mouldsevent2000Nakagaki 2000Nakagaki 2000 (Physarum polycephalum): The plasmodium finds the shortest path through a maze between two food sources.
Slime mouldsevent2008Saigusa 2008Saigusa 2008 (Physarum polycephalum): After three unfavourable pulses at fixed intervals the plasmodia slowed again at the time the next pulse would have come, which the authors read as anticipation.
Slime mouldsevent2011Latty 2011 (decoy)Latty 2011 (decoy) (Physarum polycephalum): Preferences were transitive, but adding a third, inferior food option raised the preference for the focal one, which breaks the independence-of-irrelevant-alternatives rule.
Slime mouldsevent2011Latty 2011 (speed)Latty 2011 (speed) (Physarum polycephalum): Individuals that chose fast made more costly errors (picked the worst of three foods), a speed-accuracy trade-off that light and hunger changed.
Slime mouldsevent2012Reid 2012Reid 2012 (Physarum polycephalum): The plasmodium avoids areas it has already explored, and so escapes a U-shaped trap on its way to food; the authors call this an externalized spatial memory.
Slime mouldsevent2015Latty 2015Latty 2015 (Physarum polycephalum): Time spent in a food patch rose after the plasmodium engulfed a high-quality food item, so patch leaving followed a rule based on experience inside the patch (a second slime mould, Didymium bahiense, was tested too).
Slime mouldsevent2016Reid 2016Reid 2016 (Physarum polycephalum): In a two-armed bandit task the cell compared options, integrated repeated samples and combined reward frequency with magnitude, which the authors model as an exploit-in-proportion algorithm.
Slime mouldsevent2016Boisseau 2016Boisseau 2016 (Physarum polycephalum): Plasmodia learnt to cross bridges with quinine or caffeine when the stimulus was repeated, responded again after a pause, and the effect was specific to the stimulus.
Slime mouldsevent2016Vogel 2016Vogel 2016 tests Boisseau 2016: found. A slime mould habituated to a repellent passed the response on to naive ones by cell fusion, which had to last a certain time.
Slime mouldsevent2019Dussutour 2019Dussutour 2019 (Physarum polycephalum): Slow explorers made more accurate choices than fast explorers, and longer sampling raised accuracy: a speed-accuracy link tied to individual differences between cells.
Slime mouldsevent2019Boussard 2019Boussard 2019 tests Boisseau 2016: found. Habituation to sodium survived a dormant stage of one month; the slime mould absorbed and kept the repellent, which the authors call a circulating memory.
Slime mouldsevent2021Kramar 2021Kramar 2021 (Physarum polycephalum): Tubes of the plasmodium grow or shrink around a food source, so the food location is stored in the tube-diameter pattern and shapes later migration.
Slime mouldsevent2023Yoneoka 2023Yoneoka 2023 tests Boisseau 2016: found. Ran the quinine-bridge habituation experiment "in the same manner" as the original and observed habituation; the transport network changed from tree to mesh while the cell habituated.
Slime mouldsevent2023Kippenberger 2023Kippenberger 2023 (Physarum polycephalum): In a maze of Y-branches the cell was a quick learner at the first bifurcations of finding the rewarded branch; food soaked with propofol was chosen less.
Slime mouldsevent2025Yin 2025Yin 2025 tests Latty 2011 (decoy): found. Added a third yeast species as a decoy and the slime mould shifted its preference between two others (a decoy effect), in a different species of slime mould.
Slime mouldsevent2026Kuttner 2026Kuttner 2026 (Physarum polycephalum): Spontaneous alternation in T-mazes appeared only at the shortest distance (3 mm); the authors say it may reveal spatial memory but cannot rule out the maze shape.
Amoebaeevent2014Skoge 2014Skoge 2014 (Dictyostelium discoideum): Cells kept moving toward the source of chemical waves in the back of the wave for a natural 6-minute period and increasingly reversed for longer periods, which the authors call cellular memory.
Amoebaeevent2019De 2019De 2019 (Amoeba proteus): Amoebae that moved toward an electrode while a chemical peptide was present kept moving that way for 44 minutes on average after the peptide was removed.
Amoebaeevent2021Karmakar 2021Karmakar 2021 tests Skoge 2014: found. The memory depends on the background level of the chemical: chemotaxis was best at intermediate levels and fell to almost zero at high ones.
Amoebaeevent2021van 2021van 2021 (Dictyostelium discoideum): Two memories steer movement: about 20 seconds of the last pseudopod position, and about 2 minutes of the activity of the last ten pseudopods.
Amoebaeevent2021Carrasco-Pujante 2021Carrasco-Pujante 2021 tests De 2019: found. Tracked more than 2,000 cells of three amoeba species and again found the conditioned migration pattern.
Amoebaeevent2026Mueller 2026Mueller 2026 tests Saigusa 2008: found. After four periodic blue-light pulses the amoebae slowed at the time the next pulse would have come, as in the slime mould study.
Ciliatesevent1982Csaba 1982Csaba 1982 (Tetrahymena pyriformis): A single dose of a hormone-like molecule raised cell division for as many as 500 generations and re-exposure enhanced it again, which the authors call receptor "memory".
Ciliatesevent1984Csaba 1984Csaba 1984 tests Csaba 1982: found. Spaced exposures induced the memory better than one long exposure and a foreign hormone could extinguish it (retroactive interference).
Ciliatesevent1988Wood 1988 (a)Wood 1988 (a) (Stentor coeruleus): Stentor habituates to repeated mechanical taps; the drop in response tracks a smaller receptor potential and depends mainly on action potentials.
Ciliatesevent1988Wood 1988 (b)Wood 1988 (b) tests Wood 1988 (a): found. Habituation comes from a change in the mechanoreceptor channels (the voltage-dependent part of the current became steeper), not from fewer channels.
Ciliatesevent2006Armus 2006Armus 2006 (Paramecium caudatum): Paramecia trained with electric shock on a brightness discrimination task: the authors write that learning may have occurred.
Ciliatesevent2009Mingee 2009Mingee 2009 tests Armus 2006: not found. Same laboratory: paramecia did not learn a discrete action with shock as reinforcer (sample of 40).
Ciliatesevent2019Dexter 2019Dexter 2019 tests Jennings 1906: found. Using the correct species, the authors confirmed Jennings' observations: S. roeseli shows a hierarchy of avoidance behaviours, with large differences between cells.
Ciliatesevent2022Alipour 2022Alipour 2022 tests Armus 2006: found. Used the Armus brightness-and-shock task: trained paramecia differed from controls; midazolam disrupted the learning, a microtubule drug did not.
Ciliatesevent2023Rajan 2023Rajan 2023 tests Wood 1988 (a): found. Habituation to repeated taps was seen in single cells, but as one step-like switch per cell, not the graded decline earlier studies reported.
Ciliatesevent2026Rajan 2026 (protein)Rajan 2026 (protein) tests Wood 1988 (a): found. Blocking protein synthesis sped up habituation and lengthened memory; calcium signalling and protein phosphorylation were implicated.
Ciliatesevent2026Rajan 2026 (S. stipatus)Rajan 2026 (S. stipatus) tests Wood 1988 (a): found. A newly described Stentor species also habituates to repeated taps (title and behavioural characterisation).
Fungievent2020Fukasawa 2020Fukasawa 2020 (Phanerochaete velutina): Mycelium regrew from 67% of inocula moved to fresh soil; a bait-size threshold cued the move; the authors tested the mycelium's memory of growth direction.
Fungievent2025Parise 2025Parise 2025 tests Fukasawa 2020: not found. A similar test of memory of the food direction in an ectomycorrhizal fungus found no evidence of memory.
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

Claims came first and tests came late. Of the 30 first reports, 23 were published in 2010 or later, but of the 11 independent tests only 1 appeared before 2020 (Dexter and colleagues in 2019). Part of this is simple timing: a 2025 claim has had little time to be repeated. The rest is the pattern the ledger documents: independent repeats are rare enough to count one by one.

Figure 3

Claims came first, independent tests came late: papers by decade

Left: first reports (30). Right: independent tests (11: repeats, other-species tests and one extension). 10 of the 11 independent tests are from 2020 or later. Papers from the 1950s-1970s with no abstract online could not be counted. As of 10 November 2026.

Source: Our ledger; counts by IQ Metrics

Data table
decadeFirst reportsIndependent tests
2020s910
2010s141
2000s40
1990s00
1980s20
Before 198010
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

05 Which kinds of learning are claimed, and in which organisms?

Memory is the most common claim, 13 of the 30 first reports, followed by choice with a named bias (7), habituation (5) and associative learning (5). Slime moulds hold 12 of the claims, plants 10, ciliates 4, amoebae 3 and fungi 1. Habituation is the kind most likely to be repeated: 3 of its 5 claims have an independent same-species repeat, against 0 of the 13 memory claims.

Figure 4

Slime moulds and plants carry most claims; memory and choice are the common kinds

Number of first reports (30) by organism group and kind of learning claimed. Memory is claimed 13 times, choice with a named bias 7, habituation 5 and associative learning 5. 25 of the 30 claims record behaviour (movement, growth direction, choice); 5 record gene activity or electric charge. As of 10 November 2026.

Source: Our ledger; classes defined in data/VERDICT-RULE.md

Data table
organism groupHabituationAssociative learningMemoryChoice with a named bias
Plants2341
Slime moulds1056
Amoebae0120
Ciliates2110
Fungi0010
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

“Choice with a named bias” means the authors tested a named property of a choice: the shortest path through a maze [19], a context effect when a third option is added [22], a speed-accuracy trade-off [24], risk sensitivity [9], or the exploit-or-explore problem of a two-armed bandit [28]. A plain preference for one option is not enough to enter the ledger. 5 of the 30 claims measure something other than whole-organism behaviour (gene activity in Arabidopsis and tomato, gene activity and sodium uptake after counted touches in the Venus flytrap, electric charge in the same plant, and growth of a Tetrahymena culture), and are in only because their authors call the result learning or memory [14] [15] [10] [12] [51].

Figure 5

Organism, kind of learning, verdict: how 30 claims flow

Width = number of first reports. Left: organism group. Middle: kind of learning the authors claim. Right: verdict from the written rule. Habituation is the kind most often repeated (3 of 5); memory (13 claims) and choice (7) mostly have no independent repeat. As of 10 November 2026.

Source: Our ledger; verdicts from the written rule

Data table
fromtoclaims
PlantsHabituation2
PlantsAssociative learning3
PlantsChoice with a named bias1
PlantsMemory4
Slime mouldsChoice with a named bias6
Slime mouldsMemory5
Slime mouldsHabituation1
AmoebaeMemory2
AmoebaeAssociative learning1
CiliatesHabituation2
CiliatesAssociative learning1
CiliatesMemory1
FungiMemory1
HabituationContested1
Associative learningRepeated: effect not found1
Choice with a named biasNo repeat found6
MemorySame laboratory only4
Associative learningNo repeat found2
HabituationNo repeat found1
MemoryNo repeat found6
MemoryTested in another species2
Choice with a named biasTested in another species1
HabituationRepeated: effect found3
MemoryContested1
Associative learningSame laboratory only1
Associative learningRepeated: effect found1
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

06 How many of these experiments have been repeated by someone else?

Five of the 30 first reports (17%) have been repeated by an independent laboratory in the same species, and four of those five repeats found the effect; including tests in a different species, 8 of 30 (27%) have any independent test. That leaves 22 first reports (73%) that nobody outside the original laboratory has tested in our searches, or whose only follow-up is a comment, the same laboratory or nothing.

Table 1

The verdict rule, applied mechanically to 30 first reports

Written on 10 November 2026 before any verdict was computed (data/VERDICT-RULE.md). The first test that fits wins. "Independent" = none of the repeat's authors is the first or last author of the original.

Verdict (claims)Test
A. Repeated: effect found: 4At least one independent repeat in the same species with a stated outcome, and every one found the effect.
B. Repeated: effect not found: 1At least one independent repeat in the same species found no effect.
C. Tested in another species: 3No independent same-species repeat, but an independent test in another species (outcome shown in the ledger).
D. Contested: 2None of the above, but a peer-reviewed comment, letter or re-analysis disputes the claim.
E. Same laboratory only: 5None of the above, but the same laboratory repeated or extended it.
F. No repeat found: 15None of the above in our searches. A lower bound: “we found none”, not “none exists”.

Source: IQ Metrics Research, data/VERDICT-RULE.md

Applying the rule gives 4 “repeated: found”, 1 “repeated: not found”, 3 “tested in another species”, 2 “contested”, 5 “same laboratory only” and 15 “no repeat found”. We take “effect found” and “not found” from the repeating authors’ own statement, never from our reading of their numbers, and we count an attempt that states no outcome (Ponkshe and colleagues’ pea paper [8]) as neither. A count of repeats is not a verdict on whether a finding is true: one careful repeat can outweigh several weak originals, and an original that no one has tried to repeat has simply not been tested.

Figure 6

One square per claim: 5 of 30 have been repeated by an independent laboratory

Each square is one first report of learning, memory, habituation or choice in an organism with no neurons, 1906-2026, coloured by verdict. 4 repeated with the effect found, 1 repeated with no effect, 3 tested only in another species, 2 contested by a published comment, 5 repeated or extended by the same laboratory only and 15 with no repeat found. As of 10 November 2026.

Source: Our ledger; verdicts from the written rule (data/VERDICT-RULE.md)

Data table
claimgrouplabeldetail
jennings1906Repeated: effect foundJennings 1906, StentorJennings 1906: Repeated: effect found
wood1988aRepeated: effect foundWood 1988, StentorWood 1988 (a): Repeated: effect found
armus2006Repeated: effect foundArmus 2006, ParameciumArmus 2006: Repeated: effect found
boisseau2016Repeated: effect foundBoisseau 2016, habituationBoisseau 2016: Repeated: effect found
gagliano2016Repeated: effect not foundGagliano 2016, peaGagliano 2016: Repeated: effect not found
saigusa2008Tested in another speciesSaigusa 2008, anticipationSaigusa 2008: Tested in another species
latty2011aTested in another speciesLatty 2011, decoyLatty 2011 (decoy): Tested in another species
fukasawa2020Tested in another speciesFukasawa 2020, fungusFukasawa 2020: Tested in another species
gagliano2014ContestedGagliano 2014, MimosaGagliano 2014: Contested
kramar2021ContestedKramar 2021, tubesKramar 2021: Contested
csaba1982Same laboratory onlyCsaba 1982, TetrahymenaCsaba 1982: Same laboratory only
volkov2008Same laboratory onlyVolkov 2008, flytrapVolkov 2008: Same laboratory only
skoge2014Same laboratory onlySkoge 2014, chemotaxisSkoge 2014: Same laboratory only
bohm2016Same laboratory onlyBohm 2016, Venus flytrapBöhm 2016: Same laboratory only
delafuente2019Same laboratory onlyDe la Fuente 2019, amoebaDe 2019: Same laboratory only
nakagaki2000No repeat foundNakagaki 2000, mazeNakagaki 2000: No repeat found
latty2011bNo repeat foundLatty 2011, speedLatty 2011 (speed): No repeat found
reid2012No repeat foundReid 2012, trailReid 2012: No repeat found
latty2015No repeat foundLatty 2015, patchLatty 2015: No repeat found
dener2016No repeat foundDener 2016, peaDener 2016: No repeat found
reid2016No repeat foundReid 2016, banditReid 2016: No repeat found
mittelbach2019No repeat foundMittelbach 2019, NasaMittelbach 2019: No repeat found
dussutour2019No repeat foundDussutour 2019, accuracyDussutour 2019: No repeat found
bhandawat2020No repeat foundBhandawat 2020, ArabidopsisBhandawat 2020: No repeat found
vanhaastert2021No repeat foundvan Haastert 2021, directionvan 2021: No repeat found
kippenberger2023No repeat foundKippenberger 2023, mazeKippenberger 2023: No repeat found
salvalaio2024No repeat foundSalvalaio 2024, rootsSalvalaio 2024: No repeat found
farkas2025No repeat foundFarkas 2025, tomatoFarkas 2025: No repeat found
vishton2025No repeat foundVishton 2025, MimosaVishton 2025: No repeat found
kuttner2026No repeat foundKuttner 2026, T-mazeKuttner 2026: No repeat found
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

Tests in a different species are shown apart because they test a related claim, not the same one. They were mostly positive: the slime mould anticipation result [20] was found again in Amoeba proteus [21], and the decoy effect [22] in the slime mould Physarella oblonga [23]. The exception is fungal memory of growth direction [53], which an ectomycorrhizal fungus did not show [54]. In all, 3 independent other-species tests found the effect and 1 did not.

Figure 7

Most claims since 2010 have no independent repeat yet

First reports (30) by decade of publication and verdict. 1990s: none in our ledger. Of the 23 claims published since 2010, 19 have no independent test of any kind in our searches. Recent claims have had little time to be tested. As of 10 November 2026.

Source: Our ledger; verdicts from the written rule (data/VERDICT-RULE.md)

Data table
decadeRepeated: effect foundRepeated: effect not foundTested in another speciesContestedSame laboratory onlyNo repeat foundtotal
1900s11
1980s112
2000s11114
2010s11113714
2020s1179
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

The wait for a first repeat is long and uneven: 16 years in the median, from 4 years (the pea) to 113 (Jennings’ Stentor result). The Physarum habituation claim was repeated after 7 years [29] [30]; Wood’s Stentor habituation after 35 [43] [45]; and Armus’s Paramecium learning after 16 [48] [50].

Figure 8

Years from a first report to its first independent repeat: median 16

For each of the 5 claims with an independent same-species repeat: years between the first report and the first independent repeat. Jennings reported the Stentor roeseli hierarchy in 1906; the correct species was tested again in 2019. As of 10 November 2026.

Source: Our ledger [41] [42] [43] [45] [48] [50] [29] [30] [6] [7]; years by IQ Metrics

Data table
labelvalue
Stentor roeseli, 1906 to 2019113
Stentor coeruleus, 1988 to 202335
Paramecium, 2006 to 202216
Physarum, 2016 to 20237
Pea seedlings, 2016 to 20204
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

07 Can plants learn?

Ten plant claims are in the ledger, and one of them has been independently repeated: pea associative learning (2016), where a larger, blinded replication found no effect (2020). 6 of the ten record plant movement or growth (leaf-folding, root bending, stamen movement, growth direction); 4 record gene activity, sodium uptake or electric charge, including the Venus flytrap counting study. The original authors dispute the pea replication, so we show the whole trail.

Figure 9

How four claims travelled: from first report to repeat, comment and reply

Each step in time order for four claims: pea associative learning, Mimosa habituation, Physarum habituation and the Stentor roeseli hierarchy. Steps are evenly spaced; the year is printed above each dot. The pea replication is disputed by the original authors; the Stentor repeat came 113 years after the first report. As of 10 November 2026.

First reportIndependent repeat: effect foundNot reproducedPublished commentReplySame-laboratory follow-upReview of the evidenceAttempt, no effect stated
Pea seedlings learn by association (2016)Verdict: repeated, effect not found
  1. 2016Gagliano and colleagues: pea seedlings grew toward the arm where a fan had come before light83 seedlings in a Y-maze; Scientific Reports
  2. 2020Markel: a larger, blinded replication finds no difference61 + 60 plants in the main comparison; eLife
  3. 2020Gagliano and colleagues: the replication protocol was unsuitableeLife comment
  4. 2020Markel: the set-up differences cannot explain iteLife reply
  5. 2024Ponkshe and colleagues: a second attempt lists 11 practical issuesNo effect result in the abstract
Mimosa pudica habituates to drops (2014)Verdict: contested
  1. 2014Gagliano and colleagues: leaf-folding habituates and the plants still show it a month laterOecologia
  2. 2018Biegler: no dishabituation shown, motor fatigue explains most; "remains uncertain"Oecologia comment
  3. 2018Gagliano and colleagues answer in printOecologia; no abstract in PubMed
  4. 2018Adelman: no replications of the positive conditioning results at that dateReview of plant conditioning
Slime mould habituates to bitter bridges (2016)Verdict: repeated, effect found
  1. 2016Boisseau and colleagues: Physarum learns to cross quinine and caffeine bridgesProceedings of the Royal Society B
  2. 2016Vogel and Dussutour: habituation passes between fused cellsSame laboratory
  3. 2019Boussard and colleagues: it survives one month of dormancySame laboratory
  4. 2023Yoneoka and colleagues: same quinine-bridge protocol, habituation confirmedIndependent laboratory
Stentor roeseli avoidance hierarchy (1906)Verdict: repeated, effect found
  1. 1906Jennings: a hierarchy of responses to repeated stimulationBook; described in Dexter et al. 2019
  2. 1967Reynierse and Walsh: not reproduced, with another species (S. coeruleus)As described in Dexter et al. 2019
  3. 2019Dexter and colleagues: correct species obtained, observations confirmedCurrent Biology

Source: Papers in our ledger [6] [7] [57] [58] [8] [5] [55] [61] [29] [30] [42]; steps assembled by IQ Metrics

Data table
claimstepyeartypewhat happeneddetail
Pea seedlings learn by association (2016)12016First reportGagliano and colleagues: pea seedlings grew toward the arm where a fan had come before light83 seedlings in a Y-maze; Scientific Reports
Pea seedlings learn by association (2016)22020Not reproducedMarkel: a larger, blinded replication finds no difference61 + 60 plants in the main comparison; eLife
Pea seedlings learn by association (2016)32020Published commentGagliano and colleagues: the replication protocol was unsuitableeLife comment
Pea seedlings learn by association (2016)42020ReplyMarkel: the set-up differences cannot explain iteLife reply
Pea seedlings learn by association (2016)52024Attempt, no effect statedPonkshe and colleagues: a second attempt lists 11 practical issuesNo effect result in the abstract
Mimosa pudica habituates to drops (2014)12014First reportGagliano and colleagues: leaf-folding habituates and the plants still show it a month laterOecologia
Mimosa pudica habituates to drops (2014)22018Published commentBiegler: no dishabituation shown, motor fatigue explains most; "remains uncertain"Oecologia comment
Mimosa pudica habituates to drops (2014)32018ReplyGagliano and colleagues answer in printOecologia; no abstract in PubMed
Mimosa pudica habituates to drops (2014)42018Review of the evidenceAdelman: no replications of the positive conditioning results at that dateReview of plant conditioning
Slime mould habituates to bitter bridges (2016)12016First reportBoisseau and colleagues: Physarum learns to cross quinine and caffeine bridgesProceedings of the Royal Society B
Slime mould habituates to bitter bridges (2016)22016Same-laboratory follow-upVogel and Dussutour: habituation passes between fused cellsSame laboratory
Slime mould habituates to bitter bridges (2016)32019Same-laboratory follow-upBoussard and colleagues: it survives one month of dormancySame laboratory
Slime mould habituates to bitter bridges (2016)42023Independent repeat: effect foundYoneoka and colleagues: same quinine-bridge protocol, habituation confirmedIndependent laboratory
Stentor roeseli avoidance hierarchy (1906)11906First reportJennings: a hierarchy of responses to repeated stimulationBook; described in Dexter et al. 2019
Stentor roeseli avoidance hierarchy (1906)21967Not reproducedReynierse and Walsh: not reproduced, with another species (S. coeruleus)As described in Dexter et al. 2019
Stentor roeseli avoidance hierarchy (1906)32019Independent repeat: effect foundDexter and colleagues: correct species obtained, observations confirmedCurrent Biology
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

The sensitive plant Mimosa pudica folds its leaves when touched. Gagliano and colleagues reported in 2014 that the folding habituates to repeated drops, more strongly and for longer in costly environments, and still appeared after a month [5]. Biegler answered in the same journal that the data show no dishabituation, that stimulus specificity was not tested and that motor fatigue explains most of it, concluding that “the status of habituation in Mimosa remains uncertain” [55]. We found no independent repeat of the experiment, so the verdict is “contested”. Adelman’s review of plant conditioning found the same gap in 2018 [61].

The pea experiment is the one clean test. In 2016 Gagliano and colleagues trained pea seedlings in a Y-maze, with a fan as the neutral cue, and report that the seedlings later grew toward the fan alone; the largest experiment used 83 seedlings [6]. In 2020 Markel repeated the protocol with a larger sample and blinded scoring and found no significant difference between trained and control plants (for example 61 plus 60 plants, p = 0.335) [7]. Gagliano and colleagues replied that the protocol was unsuitable for testing associative learning [57], and Markel answered that the differences between the set-ups cannot explain the lack of replication [58]. A second group attempted the experiment and reported 11 practical issues without stating an effect [8]. By our rule the claim is “repeated: not found”, and the dispute stays on the page.

The other plant claims have no independent test. Dener and colleagues report that split-root pea plants put more root growth on the variable side when mean nutrients were low, as risk-sensitivity theory predicts, and note that non-adaptive mechanisms may explain part [9]. The Venus flytrap is reported to count touches: the jasmonate pathway switches on after the second stimulus and digestive genes after more than three, with a single touch “memorized” [10]; its own laboratory followed up with a mutant that cannot count [11], and an earlier laboratory reported short-term electrical memory in the same plant [12]. Arabidopsis roots in a weak electric field overshoot and then habituate [16], and Nasa poissoniana stamens move earlier after regular visits [17]. Two claims call changes in gene activity associative learning, with music and heat in Arabidopsis [14] and with ultrasound and drought in tomato [15].

Whether these results mean that plants are “intelligent” is a dispute about definitions that this page does not enter: one group argues that regarding plants as intelligent is of real biological benefit [66], another that plants neither possess nor require consciousness [67]. Stress “priming”, where a plant exposed to heat or an attack responds faster next time, is a separate and larger literature, reviewed by Hilker and colleagues [65]; we left it out unless the authors themselves called the result learning.

08 Can slime moulds remember and decide?

Slime moulds and amoebae hold 15 of the 30 claims, and one has an independent same-species repeat: habituation of Physarum polycephalum to bitter bridges, reported in 2016 [29] and reproduced in 2023 by a second laboratory using the same quinine-bridge protocol [30]. Two more were found again in a different species, and the rest have no independent test.

Boisseau and colleagues report that the plasmodium learned to ignore quinine or caffeine when the stimulus was repeated, responded again after a pause, and showed stimulus specificity, which separates habituation from fatigue [29]. Yoneoka and colleagues ran the experiment “in the same manner” and found habituation again, while the transport tubes changed from a tree to a mesh [30]. The same laboratory as the original added that habituated slime moulds can pass the response on by fusing with naive ones [31] and that it survives a month of dormancy [32]; both are counted as same-laboratory follow-ups, not as independent repeats.

The best-known slime mould result has no independent repeat in our ledger. Nakagaki and colleagues reported in 2000 that the plasmodium finds the shortest path through a maze [19]; it is the most cited claim in the ledger, with 1,032 citations, but later maze papers mostly use a model of the organism rather than the organism, and we did not count them. Other claims: avoiding areas already explored gives an externalized spatial memory [27]; slowing down at the time of the next pulse suggests anticipation [20]; preferences shifted when a third option was added [22]; faster choices were less accurate [24] [26]. The idea that tube diameters store the location of food [33] has been met with a PNAS Letter titled “Missing evidence for memory in the monocellular slime mold” [59], whose text we could not read, so the verdict is “contested” on its title alone.

Cells of the social amoeba Dictyostelium keep moving toward the source of chemical waves in the back of the wave, which the authors call cellular memory [36], and keep two kinds of directional memory of about 20 seconds and about 2 minutes [38]. Amoeba proteus moved on in the direction learned with an electric field and a peptide for 44 minutes on average [39], a result the same laboratory extended to three species and more than 2,000 cells [40].

09 Can a single cell learn? Stentor and Paramecium

Three of the four ciliate claims were repeated by an independent laboratory and every repeat found the effect: Stentor roeseli (1906, repeated 2019), Stentor coeruleus habituation (1988, repeated 2023) and Paramecium caudatum learning (2006, repeated 2022). These are single cells, so a repeated effect there cannot involve a nerve net.

Jennings described a hierarchy of responses to repeated stimulation in Stentor roeseli in 1906. It was reported non-reproducible in 1967 using the motile species Stentor coeruleus; Dexter and colleagues obtained the correct species and confirmed Jennings’ observations in 2019, with large differences between cells [42]. Wood reported in 1988 that Stentor coeruleus habituates to repeated taps [43], through a change in the mechanoreceptor channels [44]. Rajan and colleagues repeated this in single cells in 2023 and found habituation again, but as one step-like switch per cell rather than the graded decline earlier work reported [45]; the same laboratory later found that blocking protein synthesis speeds habituation [46] and that a new Stentor species also habituates [47].

For Paramecium the record is mixed. Armus and colleagues trained paramecia with electric shock on a brightness discrimination and wrote that learning may have occurred [48]; the same laboratory then found no learning of a discrete action in 40 paramecia [49]. In 2022 an independent group used the same task and reported that trained paramecia differed from controls, that midazolam disrupted the learning and that a microtubule drug did not [50]. A 1994 test of the older tube-escape results in Stentor and Paramecium concluded that the improvement is not associative learning, because the same change occurs whatever the reinforcing stimulus [62]. The 1950s experiments of Beatrice Gelber on Paramecium and the criticisms of Donald Jensen are described in a review [60]; we did not count them because their own abstracts are not online.

Tetrahymena shows how a claim can stay in one laboratory: Csaba and colleagues reported that one dose of a hormone-like molecule raised cell division for as many as 500 generations and called it receptor “memory” [51], and followed it with nine further papers over 12 years (one counted here [52]). We found no independent repeat.

10 What is the smallest organism for which a learning result has been repeated?

Paramecium caudatum, a single cell about 300 micrometres long with no neurons, is the smallest organism in the ledger whose claimed learning was repeated by an independent laboratory with the effect found. The independent group describes the species as large, about 300 micrometres [50]; the next,Stentor coeruleus, is 1-2 mm when extended [46]. We could not read a length for Physarum plasmodia or for Stentor roeseli, so this is the smallest of the organisms for which a length is in the papers we read.

For scale, our ledger has no organism with a neuron, but the reference animal does: the nematode Caenorhabditis elegans has 302 neurons [69] and shows short-term habituation and retention of habituation lasting at least 24 hours [68]. Our studies of how much of the brain has been mapped and neurons and test results across animals put that count in context, and our news articles on what bees and ants can do and human and dog neurons show what animals with neurons have been measured doing.

11 Are the repeated results cited more than the unrepeated ones?

Not clearly: the 4 claims with an independent repeat that have a citation count have a median of 146 OpenAlex citations against 62 for the 25 others, but four papers cannot carry a comparison and a citation count is attention, not quality. Independent repeats are cited less than first reports in four of five pairs: the median of the 6 repeats with a count is 32.5, against 62 for the 29 first reports with a count, and the median comment, reply or letter has 20 [4].

Figure 10

The most cited claim, the 2000 maze study, has 1,032 citations and no independent repeat

Each dot is one paper in the ledger: publication year (across) and OpenAlex citations on 10 November 2026 (up, log scale). 45 papers with at least one citation are plotted; 5 rows have zero or no count. Citation counts measure attention, not whether the finding holds. As of 10 November 2026.

First reportIndependent testSame-laboratory follow-up

Source: Citation counts from OpenAlex [4] for the papers of our ledger

Data table
paperyearcitationsgroup
Gagliano 2014, Mimosa2014352first
Gagliano 2016, pea2016246first
Markel 2020202051indep
Ponkshe 2024202414indep
Dener 2016, pea2016104first
Bohm 2016, Venus flytrap2016181first
Iosip 2023202313follow
Volkov 2008, flytrap200856first
Volkov 2009200976follow
Bhandawat 2020, Arabidopsis202027first
Farkas 2025, tomato20256first
Salvalaio 2024, roots20245first
Mittelbach 2019, Nasa201916first
Vishton 2025, Mimosa20251first
Nakagaki 2000, maze20001032first
Saigusa 2008, anticipation2008413first
Mueller 202620261indep
Latty 2011, decoy2011160first
Yin 202520252indep
Latty 2011, speed201183first
Latty 2015, patch201533first
Dussutour 2019, accuracy201928first
Reid 2012, trail2012294first
Reid 2016, bandit2016105first
Boisseau 2016, habituation2016259first
Yoneoka 202320233indep
Vogel 2016201692follow
Boussard 2019201971follow
Kramar 2021, tubes2021104first
Kippenberger 2023, maze20239first
Skoge 2014, chemotaxis2014133first
Karmakar 2021202125follow
van Haastert 2021, direction202119first
De la Fuente 2019, amoeba201962first
Carrasco-Pujante 2021202134follow
Dexter 20192019121indep
Wood 1988, Stentor198838first
Wood 1988 (b)198855follow
Rajan 2023202369indep
Rajan 2026 (protein)20266indep
Rajan 2026 (S. stipatus)20262indep
Armus 2006, Paramecium200646first
Mingee 200920094follow
Csaba 1982, Tetrahymena198228first
Fukasawa 2020, fungus202074first
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

Figure 11

Four of five first reports are cited more than their independent repeat

OpenAlex citations on 10 November 2026 for each first report and its independent same-species repeat (5 pairs; Jennings 1906 has no count). The repeat of the quinine-bridge slime mould study has 3 citations against 259 for the original; the Stentor repeat has more than the original (69 against 38). Newer papers have had less time to be cited, and a citation is not an endorsement. As of 10 November 2026.

First reportIndependent repeat

Citations in OpenAlex

Source: OpenAlex [4]; pairs from our ledger

Data table
pairFirst reportIndependent repeat
Boisseau 2016 vs Yoneoka2593
Gagliano 2016 vs Markel24651
Gagliano 2016 vs Ponkshe 202424614
Armus 2006 vs Alipour460
Wood 1988 vs Rajan3869
Embed this chart

Paste this HTML into your page: it carries the numbers and the credit link to this figure.

Only the Stentor repeat out-cites its original. The pea replication has 51 citations against 246 for the 2016 claim, and the pea comment and reply have 20 and 21 [6] [7] [57] [58]. Newer papers have had less time to gather citations, so we show the counts and draw no conclusion about quality.

12 What does this show, and what does it not?

The ledger shows what has been published and checked, not what any organism experiences. The experiments report that cells and plants responded differently after earlier events, and for four of them an independent laboratory saw the same change again; none measures intelligence, and none is an IQ. An IQ test measures human reasoning against a norm group; the claims here are narrower, habituation, association, memory and choice. Our articles on whether intelligence is limited to IQ and the history of intelligence testing explain how far that measure goes.

Replication is a test of a result, not of its meaning. The same question was asked of famous human findings in our replication scorecard and in a news piece on whether dual n-back training raises IQ; here the method is the same and the stakes are lower. For brain tissue grown in a dish, see organoid intelligence, and for animals with neurons octopus and crow intelligence, or try the memory game and read how memory relates to IQ in people.

Cells can be tested for habituation. People can be tested for reasoning.

Our IQ test measures pattern reasoning and compares your result with adults, with a percentile.

Take the IQ test

13 Every experiment in our ledger

Table 2

Every experiment in our ledger (50 rows)

Grouped by organism, oldest first. Verdict applies to first reports only (rule in the method section); tests that follow get the outcome their authors state. n = organisms or cells in the main experiment where the text gives one clear number, "-" = not stated in what we could read. Citations = OpenAlex on 10 November 2026. The CSV adds journal, DOI, links between rows, disputes and licence. As of 10 November 2026.

Skoge 2014 [36]2014Dictyostelium discoideumMemoryFirst reportSame lab only–133Cells kept moving toward the source of chemical waves in the back of the wave for a natural 6-minute period and increasingly reversed for longer periods, which the authors call cellular memory.
De 2019 [39]2019Amoeba proteusAssociative learningFirst reportSame lab only16062Amoebae that moved toward an electrode while a chemical peptide was present kept moving that way for 44 minutes on average after the peptide was removed.
Carrasco-Pujante 2021 [40]2021Amoeba proteus, Metamoeba leningradensis, Amoeba borokensisAssociative learningSame-lab follow-upEffect found2,00034Tracked more than 2,000 cells of three amoeba species and again found the conditioned migration pattern.
Karmakar 2021 [37]2021Dictyostelium discoideumMemorySame-lab follow-upEffect found–25The memory depends on the background level of the chemical: chemotaxis was best at intermediate levels and fell to almost zero at high ones.
van 2021 [38]2021Dictyostelium discoideumMemoryFirst reportNo repeat found–19Two memories steer movement: about 20 seconds of the last pseudopod position, and about 2 minutes of the activity of the last ten pseudopods.
Mueller 2026 [21]2026Amoeba proteusMemoryOther-species testEffect found211After four periodic blue-light pulses the amoebae slowed at the time the next pulse would have come, as in the slime mould study.
Jennings 1906 [41]1906Stentor roeseliHabituationFirst reportRepeated: found––Jennings described a hierarchy of responses to repeated stimulation in the sessile ciliate Stentor roeseli, among the most complex behaviours reported for a single cell.
Csaba 1982 [51]1982Tetrahymena pyriformisMemoryFirst reportSame lab only–28A single dose of a hormone-like molecule raised cell division for as many as 500 generations and re-exposure enhanced it again, which the authors call receptor “memory”.
Csaba 1984 [52]1984Tetrahymena pyriformisMemorySame-lab follow-upEffect found––Spaced exposures induced the memory better than one long exposure and a foreign hormone could extinguish it (retroactive interference).
Wood 1988 (a) [43]1988Stentor coeruleusHabituationFirst reportRepeated: found–38Stentor habituates to repeated mechanical taps; the drop in response tracks a smaller receptor potential and depends mainly on action potentials.
Wood 1988 (b) [44]1988Stentor coeruleusHabituationSame-lab follow-upEffect found–55Habituation comes from a change in the mechanoreceptor channels (the voltage-dependent part of the current became steeper), not from fewer channels.
Armus 2006 [48]2006Paramecium caudatumAssociative learningFirst reportRepeated: found–46Paramecia trained with electric shock on a brightness discrimination task: the authors write that learning may have occurred.
Mingee 2009 [49]2009Paramecium caudatumAssociative learningSame-lab follow-upEffect not found404Same laboratory: paramecia did not learn a discrete action with shock as reinforcer (sample of 40).
Dexter 2019 [42]2019Stentor roeseliHabituationIndependent repeatEffect found–121Using the correct species, the authors confirmed Jennings’ observations: S. roeseli shows a hierarchy of avoidance behaviours, with large differences between cells.
Alipour 2022 [50]2022Paramecium caudatumAssociative learningIndependent repeatEffect found1050Used the Armus brightness-and-shock task: trained paramecia differed from controls; midazolam disrupted the learning, a microtubule drug did not.
Rajan 2023 [45]2023Stentor coeruleusHabituationIndependent repeatEffect found2269Habituation to repeated taps was seen in single cells, but as one step-like switch per cell, not the graded decline earlier studies reported.
Rajan 2026 (protein) [46]2026Stentor coeruleusHabituationIndependent extensionEffect found–6Blocking protein synthesis sped up habituation and lengthened memory; calcium signalling and protein phosphorylation were implicated.
Rajan 2026 (S. stipatus) [47]2026Stentor stipatusHabituationOther-species testEffect found–2A newly described Stentor species also habituates to repeated taps (title and behavioural characterisation).
Fukasawa 2020 [53]2020Phanerochaete velutinaMemoryFirst reportOther species18074Mycelium regrew from 67% of inocula moved to fresh soil; a bait-size threshold cued the move; the authors tested the mycelium’s memory of growth direction.
Parise 2025 [54]2025Laccaria bicolorMemoryOther-species testEffect not found–0A similar test of memory of the food direction in an ectomycorrhizal fungus found no evidence of memory.
Volkov 2008 [12]2008Dionaea muscipulaMemoryFirst reportSame lab only–56The trap sums small electrical charges that alone do not close it and shuts when a threshold is reached, which the authors call short-term electrical memory.
Volkov 2009 [13]2009Dionaea muscipulaMemorySame-lab follow-upEffect found–76Measured the charge that closes the trap: 8 microcoulombs (small trap), 9 (large trap), 4.1 at 28-36 degrees C; the cumulative effect points to short-term electrical memory.
Gagliano 2014 [5]2014Mimosa pudicaHabituationFirst reportContested–352Leaf-folding in response to repeated drops habituates, more strongly and for longer in costly environments, and the plants still showed the learned response after a month undisturbed.
Böhm 2016 [10]2016Dionaea muscipulaMemoryFirst reportSame lab only–181The trap counts touch-induced action potentials: the jasmonate pathway switches on after the second stimulus and digestive genes after more than three; a single touch is “memorized” as a subthreshold signal.
Dener 2016 [9]2016Pisum sativumChoice with a named biasFirst reportNo repeat found–104Split-root pea plants put more root growth on the variable side when mean nutrients were low and on the constant side when high, as risk-sensitivity theory predicts; the authors note non-adaptive mechanisms may explain part.
Gagliano 2016 [6]2016Pisum sativumAssociative learningFirst reportRepeated: not found83246Pea seedlings trained in a Y-maze with a fan (neutral cue) before light grew toward the fan alone, which the authors call associative learning.
Mittelbach 2019 [17]2019Nasa poissonianaMemoryFirst reportNo repeat found–16Stamens presented pollen earlier after flowers had been touched at fixed intervals; the authors report first evidence of pre-emptive stamen presentation based on remembered visit intervals.
Bhandawat 2020 [14]2020Arabidopsis thalianaAssociative learningFirst reportNo repeat found–27Seedlings conditioned with music and heat switched on heat-response genes more strongly than with music alone; the authors call this associative learning (gene expression, not movement).
Markel 2020 [7]2020Pisum sativumAssociative learningIndependent repeatEffect not found12151A replication with a larger sample and blinded scoring found no significant difference between trained and control plants.
Iosip 2023 [11]2023Dionaea muscipulaMemorySame-lab follow-upEffect found–13A Venus flytrap mutant still fires touch action potentials but cannot count them and does not enter the hunting cycle; the authors describe a calcium clock for memory and decision.
Ponkshe 2024 [8]2024Pisum sativumAssociative learningIndependent repeatOutcome not stated–14Reports 11 practical issues met while trying to replicate the pea experiment; the abstract gives no effect result.
Salvalaio 2024 [16]2024Arabidopsis thalianaHabituationFirst reportNo repeat found715Roots exposed for hours to a weak electric field overshoot and then habituate in their bending, and past exposures change the later response (hysteresis).
Farkas 2025 [15]2025Solanum lycopersicumAssociative learningFirst reportNo repeat found1686Tomato plants conditioned with ultrasound and drought used one stimulus as a predictor of the other in 3 of 4 target genes; transcript levels also showed habituation or sustained induction (gene expression, not movement).
Vishton 2025 [18]2025Mimosa pudicaMemoryFirst reportNo repeat found–1Over 15 days of two lit days then one dark day, leaf movements shifted to follow the pattern; the authors say movement tracks the number of illumination events.
Nakagaki 2000 [19]2000Physarum polycephalumChoice with a named biasFirst reportNo repeat found–1,032The plasmodium finds the shortest path through a maze between two food sources.
Saigusa 2008 [20]2008Physarum polycephalumMemoryFirst reportOther species–413After three unfavourable pulses at fixed intervals the plasmodia slowed again at the time the next pulse would have come, which the authors read as anticipation.
Latty 2011 (decoy) [22]2011Physarum polycephalumChoice with a named biasFirst reportOther species–160Preferences were transitive, but adding a third, inferior food option raised the preference for the focal one, which breaks the independence-of-irrelevant-alternatives rule.
Latty 2011 (speed) [24]2011Physarum polycephalumChoice with a named biasFirst reportNo repeat found–83Individuals that chose fast made more costly errors (picked the worst of three foods), a speed-accuracy trade-off that light and hunger changed.
Reid 2012 [27]2012Physarum polycephalumMemoryFirst reportNo repeat found–294The plasmodium avoids areas it has already explored, and so escapes a U-shaped trap on its way to food; the authors call this an externalized spatial memory.
Latty 2015 [25]2015Physarum polycephalumChoice with a named biasFirst reportNo repeat found–33Time spent in a food patch rose after the plasmodium engulfed a high-quality food item, so patch leaving followed a rule based on experience inside the patch (a second slime mould, Didymium bahiense, was tested too).
Boisseau 2016 [29]2016Physarum polycephalumHabituationFirst reportRepeated: found–259Plasmodia learnt to cross bridges with quinine or caffeine when the stimulus was repeated, responded again after a pause, and the effect was specific to the stimulus.
Reid 2016 [28]2016Physarum polycephalumChoice with a named biasFirst reportNo repeat found–105In a two-armed bandit task the cell compared options, integrated repeated samples and combined reward frequency with magnitude, which the authors model as an exploit-in-proportion algorithm.
Vogel 2016 [31]2016Physarum polycephalumHabituationSame-lab follow-upEffect found–92A slime mould habituated to a repellent passed the response on to naive ones by cell fusion, which had to last a certain time.
Boussard 2019 [32]2019Physarum polycephalumMemorySame-lab follow-upEffect found–71Habituation to sodium survived a dormant stage of one month; the slime mould absorbed and kept the repellent, which the authors call a circulating memory.
Dussutour 2019 [26]2019Physarum polycephalumChoice with a named biasFirst reportNo repeat found–28Slow explorers made more accurate choices than fast explorers, and longer sampling raised accuracy: a speed-accuracy link tied to individual differences between cells.
Kramar 2021 [33]2021Physarum polycephalumMemoryFirst reportContested–104Tubes of the plasmodium grow or shrink around a food source, so the food location is stored in the tube-diameter pattern and shapes later migration.
Kippenberger 2023 [34]2023Physarum polycephalumMemoryFirst reportNo repeat found–9In a maze of Y-branches the cell was a quick learner at the first bifurcations of finding the rewarded branch; food soaked with propofol was chosen less.
Yoneoka 2023 [30]2023Physarum polycephalumHabituationIndependent repeatEffect found–3Ran the quinine-bridge habituation experiment “in the same manner” as the original and observed habituation; the transport network changed from tree to mesh while the cell habituated.
Yin 2025 [23]2025Physarella oblongaChoice with a named biasOther-species testEffect found–2Added a third yeast species as a decoy and the slime mould shifted its preference between two others (a decoy effect), in a different species of slime mould.
Kuttner 2026 [35]2026Physarum polycephalumMemoryFirst reportNo repeat found1,2740Spontaneous alternation in T-mazes appeared only at the shortest distance (3 mm); the authors say it may reveal spatial memory but cannot rule out the maze shape.

Source: Our ledger; PubMed [2]; PubMed Central [3]; OpenAlex [4]; verdicts by IQ Metrics

Table 3

Comments, replies and reviews we used (not counted as experiments)

A comment, reply or letter has no experiment of its own, so it is not a row; a published comment that disputes a claim is what makes a verdict "Contested". Reviews were read for context only.

Biegler 2018 [55]2018CommentGagliano 2014Argues that the data show no dishabituation, that stimulus specificity was not tested and that motor fatigue explains most of the data; “the status of habituation in Mimosa remains uncertain”.18
Gagliano 2018 [56]2018ReplyGagliano 2014Reply by Gagliano and colleagues (Oecologia 2018); PubMed holds no abstract.84
Gagliano 2020 [57]2020CommentMarkel 2020The original authors say the replication protocol “was unsuitable for testing for associative learning”.20
Markel 2020 [58]2020ReplyMarkel 2020The replicator compares the two set-ups and argues the differences “are unable to explain the lack of replication”.21
Reiber 2021 [59]2021LetterKramar 2021A PNAS Letter titled “Missing evidence for memory in the monocellular slime mold” (text not readable; PubMed holds no abstract; its reference list is Kramar and Alim 2021).0
Gershman 2021 [60]2021Review or context–eLife review of single-cell learning; revisits Gelber (Paramecium, 1950s) and Jennings (Stentor, 1906).137
Adelman 2018 [61]2018Review or context–Review of classical conditioning in plants: results mixed, “no replications of positive results” at that date.22
Hinkle 1994 [62]1994Review or context–Re-test of tube-escape “learning” in Stentor and Paramecium: the same change occurs whatever the reinforcing stimulus, so it is not associative learning.24
Dussutour 2021 [63]2021Review or context–Review of habituation in slime moulds and ciliates.115
Reid 2023 [64]2023Review or context–Review of cognition in Physarum polycephalum.54
Hilker 2016 [65]2016Review or context–Review of stress priming and memory in organisms without a nervous system (the literature this ledger leaves out).615
Calvo 2020 [66]2020Review or context–“Plants are intelligent, here’s how” (Annals of Botany): one side of the plant-cognition dispute.220
Taiz 2019 [67]2019Review or context–“Plants neither possess nor require consciousness” (Trends in Plant Science): the other side.172
Rankin 1990 [68]1990Review or context–Caenorhabditis elegans as a model for learning and memory (302 neurons): the reference point.284

Source: PubMed [2]; OpenAlex [4]

Table 4

Records we read and left out (62) and why

Reasons follow the scope rules in data/VERDICT-RULE.md. Leaving a paper out says nothing about its quality. 25 older records (1956-1982) have no abstract in PubMed and no open text, so we could not read them; they are the Spirostomum, Stentor, Vorticella and Paramecium habituation and learning papers of that era.

GELBER 1956Investigations of the behavior of Paramecium aurelia. V. The effects of autogamy (nuclear reorganization)no abstract in PubMed and no open full text we could read (Gelber 1956; described in a review)
GELBER 1956Investigations of the behavior of Paramecium aurelia. III. The effect of the presence and absence of light on the occurrence of a responseno abstract in PubMed and no open full text we could read (Gelber 1956; described in a review)
JENSEN 1957Experiments on learning in parameciano abstract in PubMed and no open full text we could read (Jensen 1957; described in a review)
KATZ 1958Apparent learning in the parameciumno abstract in PubMed and no open full text we could read (Katz and Deterline 1958; described in a review)
Applewhite 1968Temperature and habituation in a protozoanno abstract in PubMed and no open full text we could read
Applewhite 1968Retetion of habituation in a protozoan improved by low temperatureno abstract in PubMed and no open full text we could read
Applewhite 1969Protozoan habituation learning after loss of macronuclei and cytoplasmno abstract in PubMed and no open full text we could read
Applewhite 1969Metallic ions and habituation in the protozoan Spirostomumno abstract in PubMed and no open full text we could read
Wood 1969Parametric studies of the response decrement produced by mechanical stimuli in the protozoan, Stentor coeruleusno abstract in PubMed and no open full text we could read (Wood 1969)
Gardner 1970Protein and RNA inhibitors and protozoan habituationno abstract in PubMed and no open full text we could read
Applewhite 1970Protein and RNA synthesis during protozoan habituation after loss of macronuclei and cytoplasmno abstract in PubMed and no open full text we could read
Gardner 1970Temperature separation of acquisition and retention in protozoan habituationno abstract in PubMed and no open full text we could read
Applewhite 1971Theory of protozoan habituationno abstract in PubMed and no open full text we could read
Applewhite 1971Similarities in protozoan and flatworm habituation behaviorno abstract in PubMed and no open full text we could read
Tushmalova 1971[Analysis of the habituation reaction in the infusoria Spirostomum ambiguum Ehrbg. as an example of elementary memory]no abstract in PubMed and no open full text we could read
Hanzel 1972Trial and error learning in paramecium: a replicationno abstract in PubMed and no open full text we could read (1972 “a replication” of trial-and-error learning in Paramecium)
Wood 1972Generalization of habituation between different receptor surfaces of Stentorno abstract in PubMed and no open full text we could read
Wood 1973Stimulus specific habituation in a protozoanno abstract in PubMed and no open full text we could read
Patterson 1973Habituation in a protozoan Vorticella convallariano abstract in PubMed and no open full text we could read
Osborn 1973The effects of vibratory and electrical stimulation on habituation in the ciliated protozoan, Spirostomum ambiguumno abstract in PubMed and no open full text we could read
Doronin 1973[Changes in intracellular potential in Spirostomum ambigum infusoria during habituation]no abstract in PubMed and no open full text we could read
Hamilton 1974Quantitative analysis of ciliary and contractile responses during habituation training in Spirostomum ambiguumno abstract in PubMed and no open full text we could read
Doronin 1974[Habituation in unicellular organisms]no abstract in PubMed and no open full text we could read
Allewell 1977A kinetic analysis of the memory of a developmental interaction. Mating interactions in Tetrahymena pyriformisno abstract in PubMed and no open full text we could read
Eisenstein 1982Habituation and sensitization in an aneural cell: some comparative and theoretical considerationsreview of habituation and sensitization in protozoa (1982); no experiment in the abstract
Csaba 1982Influence of hormone concentration and time factor on development of receptor memory in a unicellular (Tetrahymena) model systemCsaba laboratory series (same laboratory as csaba1982); one follow-up kept as a row
Csaba 1982Receptor memory in the unicellular Tetrahymena. Impact of treatment with analogous hormonesno abstract in PubMed and no open full text we could read (Csaba series)
Csaba 1983Attempt to disturb receptor memory in a unicellular (Tetrahymena) model systemCsaba laboratory series; one follow-up kept as a row
Csaba 1984Influence of prolonged life span on receptor ‘memory’ in a unicellular organism, TetrahymenaCsaba laboratory series; one follow-up kept as a row
Csaba 1986Studies into disturbing receptor ‘memory’ in a unicellular (Tetrahymena) model system: changes in the imprinting potential on exposure to combinations of related and unrelated hormonesCsaba laboratory series; one follow-up kept as a row
Csaba 1987Hormone evolution studies: multiplication promoting and imprinting (“memory”) effects of various amino acids on TetrahymenaCsaba laboratory series; one follow-up kept as a row
Köhidai 1990Persistence of receptor “memory” induced in Tetrahymena by insulin imprintingCsaba laboratory series; one follow-up kept as a row
Köhidai 1994Effects of hormones on chemotaxis in Tetrahymena: investigations on receptor memoryCsaba laboratory series; one follow-up kept as a row
Nakagaki 2001Smart behavior of true slime mold in a labyrinthauthor’s 2001 discussion of the maze result; used as the readable summary for nakagaki2000
Nakagaki 2004Obtaining multiple separate food sources: behavioural intelligence in the Physarum plasmodiumnetwork shape around food sources (Steiner tree); no learning, memory or named-bias decision
Tagkopoulos 2008Predictive behavior within microbial genetic networksin-silico evolution of regulatory networks; evolved anticipation, simulated
Dussutour 2010Amoeboid organism solves complex nutritional challengesnutrient balancing by a slime mould; no learning, memory or named-bias decision
Reid 2013Solving the Towers of Hanoi – how an amoeboid organism efficiently constructs transport networksbuilds transport networks (Towers of Hanoi); no learning, memory or named-bias decision
Volkov 2014Memory elements in the electrical network of Mimosa pudica Lmemristor (a circuit element with memory) in a plant; an electrical property, not a learning or memory claim about behaviour
Volkov 2014Memristors in the Venus flytrapmemristor in a plant; an electrical property, not a behavioural claim
Aquino 2014Memory improves precision of cell sensing in fluctuating environmentsmodel of cell sensing; no organism experiment
Iwayama 2016Decision-making ability of Physarum polycephalum enhanced by its coordinated spatiotemporal oscillatory dynamicsdecision quality in a travelling-salesman set-up vs oscillations; no named bias tested
Whiting 2016Towards a Physarum learning chipengineering of electrode grids in a slime mould network; “learning” is about conductivity, not behaviour
Meyer 2017The role of noise in self-organized decision making by the true slime mold Physarum polycephalummathematical model with a proposed experiment; no data
Reed-Guy 2017Sensitive plant (Mimosa pudica) hiding time depends on individual and stateindividual differences in leaf-closing time; no learning, memory or named-bias test
Veresoglou 2018Fungal Decision to Exploit or Explore Depends on Growth Rateno abstract in PubMed and no open full text we could read
Trinh 2019Behavioural analysis of single-cell aneural ciliate, Stentor roeseli, using machine learning approachesmachine-learning analysis of Stentor roeseli responses; the authors say they “do not conform to any particular learning model”; no learning claim
Clark 2019Unpredictable homeodynamic and ambient constraints on irrational decision making of aneural and neural foragerscommentary article (Behavioral and Brain Sciences); no experiment
Volkov 2021Cyclic voltammetry of volatile memristors in the Venus flytrap: short-term memorymemristor in a plant; an electrical property, not a behavioural claim
Murugan 2021Mechanosensation Mediates Long-Range Spatial Decision-Making in an Aneural Organismsensing of heavy masses by growth direction: a plain preference, not learning or memory
Abramson 2021Behaviorist approaches to investigating memory and learning: A primer for synthetic biology and bioengineeringprimer on learning experiments for bioengineering; context
Cheng 2022Bodily memory in slime moldno abstract in PubMed and no open full text we could read
Wang 2023Decision-Making Underlying Support-Searching in Pea Plantsa plain preference between support diameters; no learning, memory or named-bias test
Rolland 2023Behavioural changes in slime moulds over timesame-laboratory study of ageing; the abstract gives no separate learning result
Rajan 2023Studying Habituation in Stentor coeruleusvideo protocol for Stentor habituation; method, no new result
Wright 2023Cellular learning: Habituation sans neurons in a unicellular organismcommentary on Rajan et al. 2023
Cosca 2023Do environmental stimuli modify sensitive plant (Mimosa pudica L.) risk assessment?wind and sound effects on closing and reopening latency; no learning or memory test
Rivière 2023Plants sum and subtract stimuli over different timescalesstimulus integration in wheat coleoptiles (memory only in a mathematical sense); no learning claim
Paez-Espino 2024Pavlovian-Type Learning in Environmental Bacteria: Regulation of Herbicide Resistance by Arsenic in Pseudomonas putidagene co-regulation described as “akin to Pavlovian conditioning” in an evolutionary scenario: not learning within one lifetime
Dobránszki 2025Plant memory and communication of encountersperspective on epigenetic communication; no experiment
Rajan 2025A receptor-inactivation model for single-celled habituation in Stentor coeruleusmathematical model of published Stentor data; no experiment
Rahman 2026Threshold decision-making in the Venus flytrapre-reads published measurements with a mathematical framework; no experiment

Source: Europe PMC [1]; PubMed [2]; screening by IQ Metrics

14 How we built the dataset

  1. 01

    Search

    Eight Europe PMC title and abstract searches (slime moulds, ciliates and amoebae, “without neurons” wording, plants by title, plants by abstract, fungi, bacteria, cells) for papers with habituation, learning, memory, conditioning, anticipation or decision in their title or abstract: 6,205 unique records. Preprints excluded.

  2. 02

    Screen

    A written word rule kept 951 records and scripted blockers 629. We then picked records by hand from the titles, added named leads and the citation lists of the originals, and read 137 abstracts or texts. Every decision, with its reason, is in data/screen_decisions.tsv.

  3. 03

    Code

    Each kept paper became one row: organism, kind, measure, role (first report, repeat, other-species test, same-laboratory follow-up, extension), sample size where the text gives one clear number, and what the authors report. A first report’s verdict comes from the rule in data/VERDICT-RULE.md, written before any verdict was computed.

  4. 04

    Check

    A script found every typed fact (683 checks) in the saved abstract or PubMed Central full text, confirmed from the author lists that every “independent” paper shares no first or last author with its original, and confirmed that no paper is flagged retracted in OpenAlex [4]. 11 rows were read from PubMed Central full texts [3]; the rest from PubMed abstracts [2].

  5. 05

    Compute

    Counts by organism, kind and verdict; years to the first repeat; citations. All IQ Metrics calculations, scripted from the cached files.

Read every number with these in mind

  • A count of repeats is not a vote on the truth. One careful repeat can outweigh several weak originals; we report what each repeating team stated.
  • The ledger is a lower bound: 30 first reports found by our searches and leads. A paper we missed, or one without an abstract online (25 records from 1956 to 1982), is not in it. “No repeat found” means we found none.
  • “Independent” is judged from author names (no shared first or last author), not from funding or institutions; a repeat by a former colleague would count as independent.
  • Sample sizes are blank where the text we could read does not give one clear number (12 of 50 rows have one). Four papers are read from the abstract of a later paper or a summary by the same author, and say so in the CSV.
  • The “Contested” verdict for the tube-diameter study rests on the title of a PNAS Letter whose text we could not read.
  • Gene-activity and electric-charge claims are in because their authors call them learning or memory; they are not behaviour of a whole organism.
  • Citation counts change daily and measure attention, not correctness. Nothing here is medical or agricultural advice.

Sources

  1. EMBL-EBI. Europe PMC REST API: 8 title and abstract searches (6,205 records), 17 title look-ups and citation lists of 18 originals. 2026. europepmc.org/RestfulWebServiceMetadata open; abstracts read by script and cited as facts
  2. US National Library of Medicine. PubMed (NCBI E-utilities efetch): abstracts, authors and DOIs of the 137 records read. 2026. www.ncbi.nlm.nih.gov/books/NBK25501US government service; each abstract keeps its publisher's copyright and is cited as facts only
  3. US National Library of Medicine. PubMed Central full texts (NCBI E-utilities) for 11 of the papers, read by script for sample sizes and outcomes. 2026. www.ncbi.nlm.nih.gov/pmcEach article keeps its own licence; facts cited only
  4. OurResearch. OpenAlex: citation counts (cited_by_count), licences and retraction flags for the papers in the ledger, on 10 November 2026. 2026. openalex.orgCC0 (metadata)
  5. Oecologia. Gagliano et al. (2014). Experience teaches plants to learn faster and forget slower in environments where it matters. 2014. doi.org/10.1007/s00442-013-2873-7Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  6. Scientific reports. Gagliano et al. (2016). Learning by Association in Plants. 2016. doi.org/10.1038/srep38427CC BY (open access)
  7. eLife. Markel (2020). Lack of evidence for associative learning in pea plants. 2020. doi.org/10.7554/elife.57614CC BY (open access)
  8. Quarterly journal of experimental psychology (2006). Ponkshe et al. (2024). A case study of learning in plants: Lessons learned from pea plants. 2024. doi.org/10.1177/17470218231203078Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  9. Current biology : CB. Dener et al. (2016). Pea Plants Show Risk Sensitivity. 2016. doi.org/10.1016/j.cub.2016.05.008Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  10. Current biology : CB. Böhm et al. (2016). The Venus Flytrap Dionaea muscipula Counts Prey-Induced Action Potentials to Induce Sodium Uptake. 2016. doi.org/10.1016/j.cub.2015.11.057CC BY-NC-ND: facts cited, no text, table or figure copied
  11. Current biology : CB. Iosip et al. (2023). DYSCALCULIA, a Venus flytrap mutant without the ability to count action potentials. 2023. doi.org/10.1016/j.cub.2022.12.058Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  12. Plant signaling & behavior. Volkov et al. (2008). Plant electrical memory. 2008. doi.org/10.4161/psb.3.7.5684Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  13. Bioelectrochemistry (Amsterdam, Netherlands). Volkov et al. (2009). Electrical memory in Venus flytrap. 2009. doi.org/10.1016/j.bioelechem.2009.03.005Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  14. Communicative & integrative biology. Bhandawat et al. (2020). Sound as a stimulus in associative learning for heat stress in Arabidopsis. 2020. doi.org/10.1080/19420889.2020.1713426CC BY (open access)
  15. Plant signaling & behavior. Farkas et al. (2025). Short-term transcriptional memory and association-forming ability of tomato plants in response to ultrasound and drought stress stimuli. 2025. doi.org/10.1080/15592324.2025.2556982CC BY (open access)
  16. Plant physiology. Salvalaio and Sena (2024). Long-term root electrotropism reveals habituation and hysteresis. 2024. doi.org/10.1093/plphys/kiad686CC BY (open access)
  17. Plant signaling & behavior. Mittelbach et al. (2019). Flowers anticipate revisits of pollinators by learning from previously experienced visitation intervals. 2019. doi.org/10.1080/15592324.2019.1595320Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  18. Cognitive science. Vishton and Bartosh (2025). Can Mimosa pudica Plants Enumerate Light Exposure Events?. 2025. doi.org/10.1111/cogs.70161CC BY (open access)
  19. Nature. Nakagaki et al. (2000). Maze-solving by an amoeboid organism. 2000. doi.org/10.1038/35035159CC BY-NC-ND: facts cited, no text, table or figure copied
  20. Physical review letters. Saigusa et al. (2008). Amoebae anticipate periodic events. 2008. doi.org/10.1103/physrevlett.100.018101Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  21. Scientific reports. Mueller et al. (2026). Anticipation of periodic events influences cell motility in amoeba proteus. 2026. doi.org/10.1038/s41598-026-37298-0CC BY (open access)
  22. Proceedings. Biological sciences. Latty and Beekman (2011). Irrational decision-making in an amoeboid organism: transitivity and context-dependent preferences. 2011. doi.org/10.1098/rspb.2010.1045Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  23. Journal of basic microbiology. Yin et al. (2025). Context-Dependent Food Preferences and Comparative Decision-Making in Slime Mold Physarella oblonga. 2025. doi.org/10.1002/jobm.70023Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  24. Proceedings. Biological sciences. Latty and Beekman (2011). Speed-accuracy trade-offs during foraging decisions in the acellular slime mould Physarum polycephalum. 2011. doi.org/10.1098/rspb.2010.1624Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  25. The Journal of experimental biology. Latty and Beekman (2015). Slime moulds use heuristics based on within-patch experience to decide when to leave. 2015. doi.org/10.1242/jeb.116533Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  26. Proceedings. Biological sciences. Dussutour et al. (2019). Phenotypic variability predicts decision accuracy in unicellular organisms. 2019. doi.org/10.1098/rspb.2018.2825Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  27. Proceedings of the National Academy of Sciences of the United States of America. Reid et al. (2012). Slime mold uses an externalized spatial "memory" to navigate in complex environments. 2012. doi.org/10.1073/pnas.1215037109Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  28. Journal of the Royal Society, Interface. Reid et al. (2016). Decision-making without a brain: how an amoeboid organism solves the two-armed bandit. 2016. doi.org/10.1098/rsif.2016.0030CC BY (open access)
  29. Proceedings. Biological sciences. Boisseau et al. (2016). Habituation in non-neural organisms: evidence from slime moulds. 2016. doi.org/10.1098/rspb.2016.0446Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  30. Frontiers in cell and developmental biology. Yoneoka and Takamatsu (2023). Relation between learning process and morphology of transport tube network in plasmodium of Physarum polycephalum. 2023. doi.org/10.3389/fcell.2023.1249165CC BY (open access)
  31. Proceedings. Biological sciences. Vogel and Dussutour (2016). Direct transfer of learned behaviour via cell fusion in non-neural organisms. 2016. doi.org/10.1098/rspb.2016.2382Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  32. Philosophical transactions of the Royal Society of London. Series B, Biological sciences. Boussard et al. (2019). Memory inception and preservation in slime moulds: the quest for a common mechanism. 2019. doi.org/10.1098/rstb.2018.0368Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  33. Proceedings of the National Academy of Sciences of the United States of America. Kramar and Alim (2021). Encoding memory in tube diameter hierarchy of living flow network. 2021. doi.org/10.1073/pnas.2007815118Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  34. International journal of molecular sciences. Kippenberger et al. (2023). Learning in the Single-Cell Organism Physarum polycephalum: Effect of Propofol. 2023. doi.org/10.3390/ijms24076287CC BY (open access)
  35. Protist. Kuttner and Freiberg (2026). Inducing chirality in slime mold – spontaneous alternation behavior may reveal spatial memory in Physarum polycephalum. 2026. doi.org/10.1016/j.protis.2026.126156CC BY (open access)
  36. Proceedings of the National Academy of Sciences of the United States of America. Skoge et al. (2014). Cellular memory in eukaryotic chemotaxis. 2014. doi.org/10.1073/pnas.1412197111Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  37. Physical review. E. Karmakar et al. (2021). Cellular memory in eukaryotic chemotaxis depends on the background chemoattractant concentration. 2021. doi.org/10.1103/physreve.103.012402Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  38. PloS one. van (2021). Short- and long-term memory of moving amoeboid cells. 2021. doi.org/10.1371/journal.pone.0246345CC BY (open access)
  39. Nature communications. De et al. (2019). Evidence of conditioned behavior in amoebae. 2019. doi.org/10.1038/s41467-019-11677-wCC BY (open access)
  40. Frontiers in microbiology. Carrasco-Pujante et al. (2021). Associative Conditioning Is a Robust Systemic Behavior in Unicellular Organisms: An Interspecies Comparison. 2021. doi.org/10.3389/fmicb.2021.707086CC BY (open access)
  41. Columbia University Press. Jennings (1906). Behavior of the Lower Organisms. Columbia University Press (described in the abstract of Dexter et al. 2019). 1906. doi.org/10.1016/j.cub.2019.10.059Public domain (1906 book; known only from the 2019 repeat's abstract)
  42. Current biology : CB. Dexter et al. (2019). A Complex Hierarchy of Avoidance Behaviors in a Single-Cell Eukaryote. 2019. doi.org/10.1016/j.cub.2019.10.059Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  43. The Journal of neuroscience : the official journal of the Society for Neuroscience. Wood (1988). Habituation in Stentor: a response-dependent process. 1988. doi.org/10.1523/jneurosci.08-07-02248.1988CC BY-NC-SA: facts cited, no text, table or figure copied
  44. The Journal of neuroscience : the official journal of the Society for Neuroscience. Wood (1988). Habituation in Stentor: produced by mechanoreceptor channel modification. 1988. doi.org/10.1523/jneurosci.08-07-02254.1988CC BY-NC-SA: facts cited, no text, table or figure copied
  45. Current biology : CB. Rajan et al. (2023). Single-cell analysis of habituation in Stentor coeruleus. 2023. doi.org/10.1016/j.cub.2022.11.010CC BY (open access)
  46. Current biology : CB. Rajan et al. (2026). Molecular pathways for learning in the single-cell Stentor coeruleus. 2026. doi.org/10.1016/j.cub.2026.03.080CC BY (open access)
  47. Scientific reports. Rajan et al. (2026). Stentor stipatus is a new unicellular species that demonstrates habituation and unique phototaxis. 2026. doi.org/10.1038/s41598-026-40277-0CC BY-NC-ND: facts cited, no text, table or figure copied
  48. Psychological reports. Armus et al. (2006). Discrimination learning and extinction in paramecia (P. caudatum). 2006. doi.org/10.2466/pr0.98.3.705-711Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  49. Psychological reports. Mingee and Armus (2009). Unsuccessful reinforcement of a discrete action in paramecia, P. caudatum. 2009. doi.org/10.2466/pr0.105.2.533-538Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  50. Basic and clinical neuroscience. Alipour et al. (2022). Microtubule Disruption Without Learning Impairment in the Unicellular Organism, Paramecium: Implications for Information Processing in Microtubules. 2022. doi.org/10.32598/bcn.2021.2462.1CC BY-NC: facts cited, no text, table or figure copied
  51. Experimental cell biology. Csaba et al. (1982). Development and persistence of receptor 'memory' in a unicellular model system. 1982. doi.org/10.1159/000163158Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  52. Experimental cell biology. Csaba et al. (1984). Receptor 'memory' in Tetrahymena: does it satisfy the general criteria of memory? An experimental study on induction and extinction by retroactive interference in a unicellular organism. 1984. pubmed.ncbi.nlm.nih.gov/6090239Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  53. The ISME journal. Fukasawa et al. (2020). Ecological memory and relocation decisions in fungal mycelial networks: responses to quantity and location of new resources. 2020. doi.org/10.1038/s41396-019-0536-3CC BY (open access)
  54. Communicative & integrative biology. Parise et al. (2025). An experimental approach to study foraging memory in ectomycorrhizal mycelium. 2025. doi.org/10.1080/19420889.2025.2580130CC BY (open access)
  55. Oecologia. Biegler (2018). Insufficient evidence for habituation in Mimosa pudica. Response to Gagliano et al. (2014). 2018. doi.org/10.1007/s00442-017-4012-3Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  56. Oecologia. Gagliano et al. (2018). Plants learn and remember: lets get used to it. 2018. doi.org/10.1007/s00442-017-4029-7Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  57. eLife. Gagliano et al. (2020). Comment on 'Lack of evidence for associative learning in pea plants'. 2020. doi.org/10.7554/elife.61141CC BY (open access)
  58. eLife. Markel (2020). Response to comment on 'Lack of evidence for associative learning in pea plants'. 2020. doi.org/10.7554/elife.61689CC BY (open access)
  59. Proceedings of the National Academy of Sciences of the United States of America. Reiber and Nogueira (2021). Missing evidence for memory in the monocellular slime mold. 2021. doi.org/10.1073/pnas.2105928118Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  60. eLife. Gershman et al. (2021). Reconsidering the evidence for learning in single cells. 2021. doi.org/10.7554/elife.61907CC BY (open access)
  61. Perspectives on behavior science. Adelman (2018). On the Conditioning of Plants: A Review of Experimental Evidence. 2018. doi.org/10.1007/s40614-018-0173-6Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  62. Behavioral neuroscience. Hinkle and Wood (1994). Is tube-escape learning by protozoa associative learning?. 1994. doi.org/10.1037//0735-7044.108.1.94Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  63. Biochemical and biophysical research communications. Dussutour (2021). Learning in single cell organisms. 2021. doi.org/10.1016/j.bbrc.2021.02.018Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  64. Animal cognition. Reid (2023). Thoughts from the forest floor: a review of cognition in the slime mould Physarum polycephalum. 2023. doi.org/10.1007/s10071-023-01782-1CC BY (open access)
  65. Biological reviews of the Cambridge Philosophical Society. Hilker et al. (2016). Priming and memory of stress responses in organisms lacking a nervous system. 2016. doi.org/10.1111/brv.12215Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  66. Annals of botany. Calvo et al. (2020). Plants are intelligent, here's how. 2020. doi.org/10.1093/aob/mcz155Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  67. Trends in plant science. Taiz et al. (2019). Plants Neither Possess nor Require Consciousness. 2019. doi.org/10.1016/j.tplants.2019.05.008Publisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  68. Behavioural brain research. Rankin et al. (1990). Caenorhabditis elegans: a new model system for the study of learning and memory. 1990. doi.org/10.1016/0166-4328(90)90074-oPublisher copyright: abstract or open text read; facts cited, no text, table or figure copied
  69. Philosophical Transactions of the Royal Society B. White et al. (1986). The structure of the nervous system of the nematode Caenorhabditis elegans. Philosophical Transactions of the Royal Society B (302 neurons; via IQ Metrics Research, How much of the brain have we mapped). 1986. doi.org/10.1098/rstb.1986.0056Publisher copyright: count cited as a fact

Version history

  • 2026.1110 November 2026First release. Next update due May 2027 (searches re-run every six months, since new slime mould, Stentor and plant papers appear several times a year); sooner if a replication of a first report is published or a paper in the ledger is corrected or retracted.
Leon Falkner

Leon Falkner, Data & Research Analyst. Leon owns the arithmetic on IQ Metrics: the percentile tables, the score converter and the figures in everyone else's drafts. On the research desk Leon pulls the data, builds each index and writes the method. More about Leon.

Questions

Can plants learn?

Some experiments say so, but only one of 10 plant claims was independently repeated, and it found no effect. Gagliano and colleagues reported associative learning in pea seedlings in 2016; Markel's larger, blinded replication in 2020 found no difference. The original authors dispute the replication's protocol.

Can a slime mould learn?

According to its authors, yes, in one case that was repeated: Physarum polycephalum habituated to quinine and caffeine bridges in 2016, and a second laboratory reproduced it in 2023. Of 12 slime mould claims, only that one has an independent same-species repeat.

Are plants intelligent?

That is a dispute about definitions, and our ledger takes no side. It counts experiments: 10 plant claims of learning or memory, 1 independently repeated. One group argues plants are intelligent (Calvo et al. 2020); another that they neither possess nor require consciousness (Taiz et al. 2019).

Do plants have memory?

Authors report several kinds of plant memory, such as a Venus flytrap that counts touches over seconds and a Mimosa that folds less after repeated drops, but 9 of 10 plant claims have no independent repeat. Stress priming, a separate field, is not counted here.

Can an organism without a brain learn?

The authors of 30 experiments report habituation, association, memory or choice in organisms with no neurons; 5 were independently repeated in the same species and 4 of those repeats found the effect, in Stentor, Paramecium and Physarum.

What is the smallest organism that has been shown to learn?

In our ledger, Paramecium caudatum, a single cell about 300 micrometres long with no neurons. Its 2006 learning result was repeated by an independent group in 2022 with the effect found. The authors of the first report wrote that learning "may have occurred".

Has the pea plant learning experiment been repeated?

Yes, once with a stated result: Markel's 2020 replication, with a larger sample and blinded scoring, found no significant difference (61 plus 60 plants, p = 0.335). Gagliano and colleagues replied that the protocol was unsuitable. A second attempt in 2024 listed 11 practical issues.

Can a single cell learn?

The authors of 7 single-cell claims say so; three of the four ciliate claims were independently repeated with the effect found: Stentor roeseli (1906, repeated 2019), Stentor coeruleus (1988, repeated 2023) and Paramecium caudatum (2006, repeated 2022).

Do slime moulds have a memory?

Reid and colleagues reported in 2012 that Physarum avoids areas it has already explored, an externalized spatial memory that lets it escape a U-shaped trap. We found no independent repeat of that experiment, and another 2026 T-maze study could not tell memory from maze shape.

How long does it take to repeat a learning experiment without neurons?

A median of 16 years from the first report to the first independent repeat, across 5 claims: 4 years for the pea, 7 for Physarum habituation, 16 for Paramecium, 35 for Stentor coeruleus and 113 for Stentor roeseli.

Use the data

Share it, cite it, download it

You may reuse everything on this page under CC BY 4.0: in articles, reports, lessons and posts. The one condition is credit to IQ Metrics with a link to this page.

Cite this study

Version 2026.11. Citing the version lets readers find the exact numbers.

Archived copy with a DOI: 10.5281/zenodo.24098409 (report PDF and data on Zenodo).

Falkner, L. (2026). Can plants and slime moulds learn? Every experiment, and whether it was repeated (Version 2026.11) [Data set]. IQ Metrics Research. https://iqmetrics.org/research/can-plants-and-slime-moulds-learn/ (DOI: 10.5281/zenodo.24098409)

Embed the key numbers

Paste this into an article or post. It lists the main numbers and credits the source.

Get the data

One row per published experiment (50 rows): 30 first reports and the repeats, other-species tests, same-laboratory follow-ups and extensions that tested them. Organism, kind of learning claimed, what was recorded, sample size where stated, what the authors report, outcome of each test, verdict of each first report by a written rule, links between rows, disputes, OpenAlex citations on 10 November 2026 and licence. Empty = not stated in the text we could read.

Download CSV
Data dictionary
ColumnWhat it holds
claim_id, study, year, journal, doiOur id, first author and year, publication year, journal and DOI
organism_group, organismPlants, Slime moulds, Amoebae, Ciliates or Fungi; the species
kind_of_learninghabituation, associative, memory or decision (choice with a named bias); definitions in the method section
measurebehaviour (movement, growth direction, choice) or molecular (gene activity, electric charge, culture growth)
role, tests_claimFirst report, repeat (same species), other-species test, same-lab follow-up or extension; the claim id it tests
independent_of_original, outcome_of_testWhether no author is the first or last author of the original; found / not found / not stated
n, n_whatOrganisms or cells in the main experiment where the text gives one clear number, and what is counted
what_the_authors_reportOur paraphrase of the result as the authors state it
verdict_code, verdictA-F for first reports only (rule in the method section): repeated found, repeated not found, other species, contested, same laboratory only, no repeat found
independent_repeats, other_species_tests, same_lab_followups, comments_or_letters, disputeIds of the rows and responses linked to a first report; the dispute, if any
openalex_citations_2026_11_10, paper_licenceOpenAlex cited_by_count on 10 November 2026; the licence OpenAlex records for the paper (facts only are cited)
text_read, noteWhether we read the abstract, the full text or a later paper's abstract; our note

Share this finding

A finished image and a ready message. The image carries the source line, so the credit travels with it.

Each network gets a version sized for it unless you edit the message.

Press kit

For journalists, researchers and teachers

Everything on this page may be quoted, charted and republished with credit. No embargo, no permission needed, no sign-up. If you need a number that is not here, ask.

Summary you can paste

Plain news style. Edit freely; keep the credit.

IQ Metrics Research built a ledger of 50 experiments (1906-2026) that claim learning, memory, habituation or choice in organisms with no neurons, and scored each of the 30 first reports with a rule written before the outcomes were computed. 5 were repeated by an independent laboratory in the same species; 4 of those repeats found the effect, and the one plant repeat, on pea seedlings, did not. 22 of 30 first reports have no independent test of any kind. The median wait for a first repeat is 16 years. Source: IQ Metrics Research, https://iqmetrics.org/research/can-plants-and-slime-moulds-learn/

Angles for a story

  • 5 of 30 claims of learning without neurons were repeated, and 4 of the 5 repeats held (Figure 6)
  • The pea plant that did not learn: the one independent plant test, the dispute and the reply (Figure 9)
  • A cell 300 micrometres long is the smallest organism whose learning has been independently repeated (the section)
  • 113 years: Jennings’ 1906 Stentor result confirmed in 2019 (Figure 8)
  • The most cited slime mould claim, the 2000 maze, has no independent repeat (Figure 10)
  • Independent tests arrived late: 10 of 11 are from 2020 or later (Figure 3)

Quotes you can use

Attributable as written, with the name and title. For a different angle or a quote on a specific figure, ask through the contact page.

  • The honest headline is not that plants can learn or cannot. It is that most of these experiments have never been tried by anyone else, and the ones that have been tried mostly held up, except in the pea.

    Leon Falkner, Data & Research Analyst, IQ Metrics Research
  • The Stentor result took 113 years to be repeated with the right species. That is not a failure of the cell. It is how slowly a replication gets funded.

    Leon Falkner, Data & Research Analyst, IQ Metrics Research
  • We used the words the authors used, responded, habituated, grew toward. We did not decide who is right in the plant dispute; we counted who tried the experiment again.

    Leon Falkner, Data & Research Analyst, IQ Metrics Research

Files

  • CSV of every experiment: study, year, organism, kind, role, link to the claim tested, outcome, n, verdict, disputes, citations, licence (CC BY 4.0)
  • Every figure as a print-ready PNG: use the Image button under each figure
  • Embed code and data table under every figure
  • Share images for social media in three sizes, with the credit line built in
  • Citations in APA, MLA, Chicago, BibTeX and plain text

Terms, contact and corrections

Credit: “IQ Metrics Research, Can plants and slime moulds learn? (2026)” with a link to this page. On air or in print, “IQ Metrics” and the page address are enough.

What it covers: published experiments on organisms with no neurons that claim habituation, associative learning, memory or choice with a named bias, found by our Europe PMC searches and leads; reviews and comments are context only.

Contact: the research team, through the contact page. Corrections are made on this page with a dated note in the version history.

Keep reading

Related research