1. Plant Biology
Download icon

Response to comment on 'Lack of evidence for associative learning in pea plants'

  1. Kasey Markel  Is a corresponding author
  1. Department of Plant Biology, University of California, Davis, United States
  • Cited 2
  • Views 738
  • Annotations
Cite this article as: eLife 2020;9:e61689 doi: 10.7554/eLife.61689


In 2016 Gagliano et al. reported evidence for associative learning in plants (Gagliano et al., 2016). A subsequent attempt to replicate this finding by the present author was not successful (Markel, 2020). Gagliano et al. attribute this lack of replication to differences in the experimental set-ups used in the original work and the replication attempt (Gagliano et al., 2020). Here, based on a comparison of the two set-ups, I argue that these differences are unable to explain the lack of replication in Markel, 2020.


If reproducible, the evidence for associative learning in plants reported in Gagliano et al., 2016 would require scientists to rethink the evolutionary history of learning because associative learning has only been consistently reported in metazoa. While learning does not imply phenomenological consciousness (Nagel, 1974), its possible discovery in plants nonetheless raises interesting questions about philosophy of mind (Bronfman et al., 2016), in addition to questions regarding the molecular mechanisms and evolutionary history of learning (Ginsburg and Jablonka, 2010).

The primary criticism of my replication attempt (Markel, 2020) is that light was not shown to be a reliable unconditioned stimulus (Gagliano et al., 2020). It is true that the plants in Markel, 2020 did not always grow towards the last presentation of light, which reduced the power to detect associative learning. However, Gagliano et al., 2016 does not emphasize how surprising it is that plants would always grow towards a one hour presentation of light preceded by presentations in the opposite direction. Most experiments with phototropism involve light exposure from only one side and for substantially longer periods of time (3 hr in Schumacher et al., 2018 and Haga and Kimura, 2019 3–6 hr in Sullivan et al., 2016 or 24 hr in Goyal et al., 2016). Relatively few experiments have been reported on phototropism in etiolated Pisum sativum seedlings. However, in wild-type etiolated Arabidopsis thaliana seedlings, phototropic bending is barely initiated after 1 hr of exposure (Sullivan et al., 2016). Moreover, circumnutation means that it is unlikely that phototropic bending would result in the growth of Pisum sativum into the predicted arm of the maze many hours later.

Furthermore, in phototropism experiments blue light illumination is generally presented horizontally to maximize the phototropic bending, but in the Y-maze configuration the angle of illumination comes primarily from above: if the LEDs are attached on the outside edges of the Y-maze, as in Markel, 2020, the angle of the light at soil level is less than 40° from vertical. If the lights are attached in the center of each arm of the Y-maze, the angle is less than 25° from vertical. This oblique light is unlikely to cause a phototropic response as rapid or as strong as that caused by horizontal light (see, for example, figure 1 in Sullivan et al., 2016). For technical reasons, neither Gagliano et al., 2016 nor Markel, 2020 measured the rate of phototropic bending.

Gagliano et al. mention that they also encountered conditions in which light was not an effective unconditioned stimulus, citing specifically their second experiment with various circadian phases. However, Markel, 2020 was performed in the same circadian phase as their first experiment, so circadian phase is not a candidate explanation for the less consistent phototropic growth.

Gagliano et al. also mention that their plants were ~20 cm apart (a detail not included in the 2016 paper; moreover, in supplementary video 1 for this paper the spacing between the plants appears to be less than 20 cm, though greater than the compact spacing used in Markel, 2020). Regardless, the LED within each Y-maze provides the vast majority of the light to the seedlings, as can be seen from the internal illumination of the 'dark' arms of the Y-mazes in all panels of Figure 1. Moreover, because all the lights were on simultaneously, the higher level of background light in Markel, 2020 was only present when the larger amount of light from Y-maze internal reflection was also present.

I have used a technique called histogram matching (Chang and Wong, 1978) to compare images of the experimental chamber in Gagliano et al., 2016 (using a frame extracted from supplementary video 1; Figure 1A) and Markel, 2020 (using a photograph; Figure 1B). This technique adjusts the intensity histograms of images (the distribution of pixel intensity from 0 to 255 in the case of 8-bit images such as these) in a way that allows relative comparisons of brightness, contrast, and intensity to be made between images acquired with different settings. This technique is well suited to this particular case because there is a common standard of brightness (LEDs emitting 14 μmol m−2 s−1 at wavelengths between 430 and 505 nm). The unedited images in Figure 1A and B show a large apparent difference in background brightness. However, histogram-matched images in Figure 1C and D suggest that most of the apparent difference in brightness between the two studies is due to differences in settings and hardware of the cameras used to capture the images.

Comparison of background light levels in Gagliano et al., 2016 (left) and Markel, 2020 (right).

The chambers in both experiments were lit only by blue lights within the Y-mazes; the only other light in the experiments came from the red headlamps worn by the experimenters. (A) Unedited still frame extracted from supplementary video 1 in Gagliano et al., 2016. (B) Unedited image of the Y-mazes used in Markel, 2020. (C) Panel A after histogram matching with panel B. The irregular rectangles are caused by the compression algorithm used in the original .mov file, and can be seen by adjusting the brightness and contrast on any frame extracted from that video. (D) Panel B after histogram matching with panel A. The histogram for each panel is shown in the bottom right corner; all histograms are 300 × 240 pixels, RGB, 281K; all images were processed as. png files for lossless compression. Image analysis was performed using the Fiji distribution of ImageJ Version 2.0.0-rc-69/1.52 n, Build: 269a0ad53f. The HistogramMatcher script is from the Fiji project CorrectBleach. This script was provided by Stack Overflow user Jan Eglinger, and is available in Supplementary file 1.

Nevertheless, the closer proximity of plants used in the replication attempt (Markel, 2020) resulted in a higher level of background light. The chambers used in the replication attempt were also smaller than those used in Gagliano et al., 2016, but all were maintained at the same temperature and humidity conditions. The difference in chamber size is unlikely to be important except insofar as it resulted in changes to plant spacing. Despite considerable effort to match the experimental details of the 2016 experiment, the replication attempt did not find evidence for associative learning in pea plants. Of course, this does not rule out the existence of such learning, and I sincerely hope that future research demonstrates the phenomenon to be reproducible.


  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11

Article and author information

Author details

  1. Kasey Markel

    Department of Plant Biology, University of California, Davis, Davis, United States
    Conceptualization, Software, Validation, Methodology, Writing - original draft, Writing - review and editing
    For correspondence
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8285-3888


No external funding was received for this work.

Senior Editor

  1. Christian S Hardtke, University of Lausanne, Switzerland

Reviewing Editor

  1. Daeyeol Lee, Johns Hopkins University, United States

Publication history

  1. Received: August 3, 2020
  2. Accepted: September 3, 2020
  3. Version of Record published: September 10, 2020 (version 1)


© 2020, Markel

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.


  • 738
    Page views
  • 47
  • 2

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Plant Biology
    Michael Borg et al.
    Research Article

    Alternation between morphologically distinct haploid and diploid life forms is a defining feature of most plant and algal life cycles, yet the underlying molecular mechanisms that govern these transitions remain unclear. Here, we explore the dynamic relationship between chromatin accessibility and epigenetic modifications during life form transitions in Arabidopsis. The diploid-to-haploid life form transition is governed by the loss of H3K9me2 and DNA demethylation of transposon-associated cis-regulatory elements. This event is associated with dramatic changes in chromatin accessibility and transcriptional reprogramming. In contrast, the global loss of H3K27me3 in the haploid form shapes a chromatin accessibility landscape that is poised to re-initiate the transition back to diploid life after fertilization. Hence, distinct epigenetic reprogramming events rewire transcription through major reorganization of the regulatory epigenome to guide the alternation of generations in flowering plants.

    1. Genetics and Genomics
    2. Plant Biology
    Thierry Halter et al.
    Research Article

    Active DNA demethylation has emerged as an important regulatory process of plant and mammalian immunity. However, very little is known about the mechanisms by which active demethylation controls transcriptional immune reprogramming and disease resistance. Here, we first show that the Arabidopsis active demethylase ROS1 promotes basal resistance towards Pseudomonas syringae by antagonizing RNA-directed DNA methylation (RdDM). Furthermore, we find that ROS1 facilitates the flagellin-triggered induction of the disease resistance gene RMG1 by limiting RdDM at the 3' boundary of a remnant RC/Helitron transposable element (TE) embedded in its promoter. We further identify flagellin-responsive ROS1 putative primary targets, and show that at a subset of promoters, ROS1 erases methylation at discrete regions exhibiting WRKY transcription factors (TFs) binding. In particular, we demonstrate that ROS1 removes methylation at the orphan immune receptor RLP43 promoter, to ensure DNA binding of WRKY TFs. Finally, we show that ROS1-directed demethylation of the RMG1 and RLP43 promoters is causal for both flagellin responsiveness of these genes and for basal resistance. Overall, these findings significantly advance our understanding of how active demethylases shape transcriptional immune reprogramming to enable antibacterial resistance.