Understanding interdisciplinary perspectives of plant intelligence: Is it a matter of science, language, or subjectivity?
Evidence suggests that plants can behave intelligently by exhibiting the ability to learn, make associations between environmental cues, engage in complex decisions about resource acquisition, memorize, and adapt in flexible ways. However, plant intelligence is a disputed concept in the scientific community. Reasons for lack of consensus can be traced back to the history of Western philosophy, interpretation of terminology, and due to plants lacking neurons and a central nervous system. Plant intelligence thus constitutes a novel paradigm in the plant sciences. Therefore, the perspectives of scientists in plant-related disciplines need to be investigated in order to gain insight into the current state and future development of this concept.
This study analyzed opinions of plant intelligence held by scientists from different plant-related disciplines, including ethnobiology and other biological sciences, through an online questionnaire.
Our findings show that respondents’ personal belief systems and the frequency of taking into account other types of knowledge, such as traditional knowledge, in their own field(s) of study, were associated with their opinions of plant intelligence. Meanwhile, respondents’ professional expertise, background (discipline), or familiarity with evidence provided on plant intelligence did not affect their opinions.
This study emphasizes the influential role of scientists’ own subjective beliefs. In response, two approaches could facilitate transdisciplinary understanding among scientists: (1) effective communication designed to foster change in agreement based on presented information; and (2) holding space for an interdisciplinary dialogue where scientists can express their own subjectivities and open new opportunities for collaboration.
Paradigm shifts in worldviews and perceptions are a part of human cultural evolution [1, 2]. Human–nature relationships, whether agronomic, leisurely, experimental, or spiritual, are a reflection of people’s perception of nature (including plants), contained within their cultural and environmental domains [3–5]. Plant scientists and philosophers have proposed a new paradigm in the way plants are perceived by demonstrating the potential of plants to have agency, individuality, and even intelligence [6–8].
A relatively newly termed concept, plant intelligence, refers to “any type of intentional and flexible behavior that is beneficial and enables the organism to achieve its goal” [9]. Plant behavior can be defined as “a response to an event or environmental change during the course of an individual’s lifetime” [10]. While animals use direct movement as a behavioral response, plant behavior is characterized by changes in growth and development. Phenotypic traits in response to environmental changes can either be hardwired (inflexible responses that are genetically encoded) or soft-wired (flexible responses that are flexible and dependent on changing environments). Beneficial behaviors that are adaptive (i.e., flexible) are the soft-wired, phenotypically plastic responses, capable of integrating multiple information signals and channeling an optimal global response [11]. Notably, a simple one-to-one response to an external stimulus is not considered intelligent behavior. The author in [10] also gives a list of examples to differentiate between inflexible and flexible behaviors.
Possible examples of intelligence in plants include key features that involve the ability to problem-solve in a flexible manner, anticipate the future, store memory, learn, communicate, and ultimately be goal-oriented [11–13]. Various plant behaviors can indicate intelligence (e.g., [8, 9, 14–16]). For example, induced volatile defenses against herbivores have to be produced at the appropriate time, concentration, and amount, demonstrating a plant’s flexibility (i.e., adaptability). These behavioral responses in turn affect, and are affected, by neighboring plants, shaping plant communities [17, 18]. Another example of flexible plant behavior is the capability to manipulate the concentration of secondary metabolites and control the quality of defense provided by species-specific ants [19]. An example of the ability to anticipate the future is how plants rely on light cues to remember the exact number of warm days or daylight hours (i.e., photoperiodic control) that have passed to develop their leaves and flower [20, 21]. Moreover, carnivorous plants such as the Venus fly trap, Dionaea muscipula J.Ellis, are able to “count” and “memorize” the number of mechanical stimulations to ensure trapping success [14]. This plant species and others need to store and retrieve this information when required. There exist examples of many other ways in which plants can be phenotypically flexible to prepare for the future [22]. Furthermore, it seems that plants can also learn to make new associations through multiple cues and respond adaptively [16]. Various types of foraging behavior for sunlight and nutrients exemplify how plants can be goal-oriented [23–26]. These types of key features of intelligence are independent of taxa and of a central nervous system. These behaviors ultimately have a physiological basis which involves electrochemistry [27, 28].
To assess a plant’s intelligence is to observe the behavior from a global perspective, i.e., what the behavior represents, and ask whether it represents any of the key features of intelligence (the capacity to adapt in a flexible manner). The mechanisms and physiological basis are essential to answer “How?”.
Plant intelligence contrasts the conventional view of plants, in which they behave passively without any intention in their responses. Because the developing framework of plant intelligence is conceptually new in the plant sciences, several authors have shared their disagreements in the literature [29–32]. To name a few, these authors argue that plants do not have nerves and do not make conscious decisions, characterizing plants as intelligent is unnecessary, and that the “individuality” of a plant is difficult to identify. To briefly address the latter argument, what is considered a plant individual remains an issue of debate. According to some authors (e.g., [8]), a plant individual is at the level of the genet (i.e., coming from the same genetic source). Plant intelligence is a trait characterizing the whole plant body in pursuit of maximizing its fitness as well as multiple individuals behaving as a collective group [13, 33]. In addition, individuality can also be observed by a plant’s ability to recognize self from non-self as well as kin [34]. Scientists and philosophers who advocate for plant intelligence argue that the main reasons why the idea is controversial are due to historical perceptions about plants and their hierarchical place in relation to other organisms that still predominate in society [7], non-consensus in terminological definitions, and traditional views of the cognitive sciences [35]. Traditional cognitive theories of intelligence are built on foundations that exclude plants because they are implicitly or explicitly zoocentric [36, 37] and neurocentric [38]. The literature in support of plant intelligence advocates for the autonomy of plants in their ability to make complex decisions in an ecological and purposeful way to improve their chance of survival [6].
For clarification, these literature sources do not refer to “intelligence” as intelligent design.
Western scientific knowledge is based on rational, materialistic, reductionist, and objective methods for understanding the world [39]. One of the primary ways in which natural phenomena are observed and interpreted is based on our scientific understanding of them. However, from a non-Western cultural perspective, the ontology and epistemology of understanding natural phenomena is perceived differently, which in turn contributes to the constitution of a worldview. For example, the ontology that plants (and nature) are agents (or “persons” and not “objects”) is commonly shared in some traditional cultures and has been elaborately described in the anthropological literature [40–42]. Traditional knowledge (TK) (also known as traditional ecological knowledge or TEK) is more holistic, subjective, and qualitative than Western scientific knowledge [43]. The author in [44] has summarized TK as “a cumulative body of knowledge, practice and belief, evolving by adaptive processes and handed down through generations by cultural transmission, about the relationship of living beings (including humans) with one another and with their environment.” The author referred to “traditional” as a cultural continuity derived from historical experience [44]. Several traditional cultures believe in the autonomy of nature, treat nature as close kin and perceive nature as being alive and part of a reciprocal co-creating relationship [40–42, 45–48]. Belief in the interconnectedness of all things, such as with nature, is also shared among people who care about their spirituality [49, 50].
According to different definitions of spirituality across decades of research, authors in [51] summarized spirituality as an “aspect of human functioning, experience, and existence which concerns the transcendent.” Understanding natural phenomena through a spiritual lens is centrally an experiential quality that is related to one’s well-being, not explicit to being religious, and differs across cultures [51]. Whether from a scientific or traditional lens, they share a common goal of seeking to gain a more holistic perspective of the world.
We hypothesized that academic researchers in plant-related biological sciences who are more open to different forms of knowledge and ontologies, such as traditional knowledge systems (TKS) would likely be more acceptive of the idea of plant intelligence than those who are less open to TKS. We also speculated that the former would be more mindful of, and more inclined toward, applying sociocultural perspectives in their field(s) of study (i.e., ethnobiologists). As a result, we hypothesized that perspectives of plant intelligence will not only be influenced by scientific evidence and concepts of plant behavioral responses, but also pertain to scholars’ personal worldviews of nature, especially in relation to plants.
Because of the increasing interest in plant intelligence and its features, our study was conceived as a preliminary investigation into perceptions of plant intelligence from plant scientists in the social and biological sciences. We gathered a wide range of data with the aim of bringing together multiple and contrasting perspectives to explore major themes and associations that represent scientists’ opinions of plant intelligence. Four research questions guided this study:Which factors influence opinions of plant intelligent behavior?Do other forms of knowledge systems, such as TKS, influence opinions of plant intelligence?Can plant scientists distinguish behaviors that are indicative of plant intelligence?Is the idea of animal intelligence easier to accept than that of plant intelligence?
A semi-structured questionnaire was developed and disseminated through an online survey platform, LimeSurvey, by the Université Libre de Bruxelles (ULB), from April to June 2019. Questions included multiple choice, open-ended, dichotomous, and Likert-scale formats. Likert-scale questions had an even-numbered analogue scale (ranging from 1 to 4 or 1 to 6) with only the two ends (anchors) of the scale verbally defined, making a neutral position (e.g., option 3 in a 5-point scale) impossible. An analogue scale was preferred as a reliable measurement for subjective attitudes that cannot be measured directly (e.g., the degree of plant intelligence) [52]. An even-numbered scale was preferred as a physical measurement of answers from study participants [53]. Most questions had the option “other” to allow flexibility for respondents who had a neutral response or who preferred not to make a clear choice.
Plant scientists and graduate students whose professional background and work were related to plant sciences in any disciplinary field were invited as part of our effort to include as many individuals from a wide range of plant-related sciences. The term plant-related sciences here refers to any scientific study of plants that includes both biological and social subdisciplines, ranging from plant biology and forest ecology to ethnobotany. Target participants’ e-mails were found through systematic keywords in search engines, mainly Google Search and Google Scholar (see Appendix Table 5). A news post on the Web site of the International Laboratory of Plant Neurobiology (LINV) was posted online as an open invitation to plant scientists. Similarly, co-authors shared invitations with prospective participants through their personal networks. With every e-mail invitation, the LINV post link for the study was attached. Open invitations were also posted on social media platforms such as Twitter and Facebook. In total, more than 1500 individuals were directly contacted by e-mail to participate in the online questionnaire. The survey link was active from March 26 until May 21, 2019. By only counting e-mail invitations, the questionnaire had a participation rate of approximately 10% (n = 150). Participation success through online channels could not be estimated. Prior to distribution, the questionnaire was pretested among graduate students for feedback and further revision.
Out of a total of 150 respondents, 119 completed the questionnaire, of which 29 were graduate students. Their age ranged from 20 to 79 years (average age 44 ± 15 years). These study participants were residents of 31 different countries across Africa (n = 3), America (n = 62), Asia (n = 14), Australia (n = 7), and Europe (n = 63). Respondents’ professional fields of expertise came from a wide range of plant-related sciences, including plant ecology, plant biology, epigenetics, phylogenetics, evolutionary biology, paleobotany, invasion biology, phytochemistry, biotechnology, plant signaling and behavior, plant physiology and microbiology, plant biochemistry, horticulture, forest ecology, agronomy-related studies, silviculture, medicinal plant studies, taxonomy, botany, ethnopharmacology, and ethnobotany.
The questionnaire consisted of four parts: (a) personal background information of participants, (b) personal beliefs and associations with different forms of knowledge, (c) perspectives on plant intelligence and traditional knowledge (TK), and (d) behavioral examples of plant intelligence and analogous behaviors in animals and plants. Before the start of the questionnaire, participants were asked to read a short essay of 302 words introducing the definition and concept of plant intelligent behavior.
Background information included questions about country of residence, age, profession, expertise and research interests, and years of professional experience. A psychometric scale, named “personal perspectives,” was created to understand an individual’s personal beliefs and associations with different forms of knowledge measured through a list of 12 statements, asking respondents to choose between the options: agree, disagree, or other. These were sourced and adapted from mixed methods literature related to (1) traditional knowledge (TK), especially pertaining to indigenous knowledge, (2) individual spirituality, and (3) other miscellaneous concepts that did not have a common theme (see Table 2). The concepts behind most statements were selected for their juxtaposition with Western scientific knowledge. Next, we explored whether agreement or disagreement with these statements had any significant relationship with an individual’s acceptance of plant intelligence. We hypothesized that those who are receptive to different forms of knowing and knowledge would be more accepting of plant intelligence. The questionnaire part that dealt with perspectives on plant intelligence included questions related to participants’ current knowledge of the concept of plant intelligence, their frequency of collaborations outside the plant sciences (i.e., collaborations with the humanities), their opinions on the usefulness of including TK in their field(s) of study, and their opinions on how much TK is valued by plant scientists. The final and central part of the questionnaire contained a list of short examples of plant behaviors obtained from the literature (Table 1). Each example provided essential information for study participants to answer. Respondents were asked to rank whether they considered the plant behavior example as intelligent or not.
Respondents who did not consider a behavior intelligent were asked to explain why in an open text box.
Each example represented at least one key indicator of plant intelligence (either learning, memory, anticipatory and/or adaptive behavior). These behavioral examples can be summarized as follows (Table 1): (1) habituation and conditioned learning involving two specific plant behavioral studies that tested intelligence in plants analogous to classical animal behavioral studies; (2) seven examples of plant behavioral responses involving roots and aboveground plant parts; and (3) two examples of animal behaviors on camouflaging and foraging (Cephalopods and Macaca arctoides) and two corresponding analogous behaviors in plants (Corydalis hemidicentra Hand.-Mazz and Syngonium spp., respectively). We aimed to compare analogous animal and plant behaviors in order to observe whether intelligence in animals was more likely to be accepted than plant intelligence. Two of the seven plant behavioral examples were actions representative of mere behaviors and not intelligence. Specifically, they were titled as “mycoheterotrophic parasitic plant” and “plant defense turning herbivores into cannibals.” The first is not an indicator of intelligent behavior because the example describes a fixed behavior incapable of long-term change, while the latter is actually an experimental trick not performed by the plant itself, but by the scientist(s). Even without a human agent, the chemical emission emitted by the plant does not involve adaptive behavior nor complex decision-making [55]. Both examples do not represent features of intelligence. These considerably tricky questions were included to observe if individuals could discern the difference between plant behaviors indicative of intelligence and those that were not. In other words, we tested if individuals would “score homogeneously,” i.e., answer all the behavioral examples in the same way.
Respondents were asked to write down their fields of expertise and research interests. Those who had written words inclusive of “ethno-” or “socio-” as a prefix such as “ethnobotany” or “ethnobiology” were classified as “ethnobiologists.” The word “ethnobiologist” in this study was chosen as a general representative term and is not meant in its strict sense. Terms that did not have these prefixes but were a direct interpretation of social and cultural studies were also put into this classification. The remaining respondents were classified as pertaining to non-ethnobiologists. The purpose of this division was to distinguish those who include social and cultural aspects in their work from those who do not.
We developed a PP index for each study participant as a proxy for their acceptance of non-Western ontological knowledge and beliefs, based on the statements presented in Table 2. Respondents received a score of 1 for every statement they agreed to and a 0 for every statement they disagreed to. The cumulative sum of these scores ranged between 0 (no agreement) to 12 (full agreement). Statements whereby participants had selected “other” as an answer were not taken into account in the analysis.
Interdisciplinary opinions of plant intelligence were collected either through direct questions targeted to all participants or only ethnobiologists. We then calculated two types of respondent opinions on plant intelligence. The first was directly obtained at the beginning of the questionnaire as a direct response to the question “Do you think plants are intelligent?”; and the second was an aggregate score based on all plant behavioral examples classified into two main groups (“plants are intelligent” or “plants are not intelligent”). Unclear opinions were omitted. The second opinion was derived from a score given to each respondent which represented their tendency to respond to all plant behavioral examples in the same way, either as “plants are intelligent” or “plants are not intelligent.” This score was calculated based on the responses of all Likert-scale questions indicative of plant intelligent behavior. The scores were first normalized due to uneven Likert-scale measurements (some questions had a scale from 1 to 6 while others ranged from 1 to 4). The minimum scale referred to “not an intelligent behavior” and the maximum scale “a very intelligent behavior.” Each response was given equal weight, summed, and then divided by the total maximum score (32). Each respondent was then given a final score ranging from 0 to 1. The median score of 0.68 was used as a threshold to divide the upper and lower half respondents into two groups, those who favored plant intelligence versus those who did not. These two opinions of plant intelligence were then compared to calculate the proportion of subjects whose opinions had remained the same or changed.
All data were analyzed using nonparametric tests, including Kruskal–Wallis, chi-square (χ2), Cramer’s V, and two-proportion Z test. Ordinal data were tested for significance of associations (chi-squared test). Contingency tables, where cells did not meet the chi-square test’s assumptions [62], were corrected with a Yates correction for continuity to reduce the risk of type II errors in order to prevent an overestimation of statistical significance. The standardized residuals of each cell in chi-squared tests were then observed to explore which cells contributed most to the chi-square value. A strength test, the Cramer’s V test, displayed how strong the association between the groups was, serving as an equivalent to the Pearson’s correlation test. To understand the relationships between cells, the two-proportion Z test was used. Z tests were also corrected for multiple testing by using the Yates correction method. Furthermore, Cronbach’s alpha was estimated for groups of Likert-scale questions with different items (all the behavioral examples presented) to test how internally consistent responses would be if participants repeated the questions, i.e., a coefficient of how reliable their responses were [63]. The percent distribution, Likert mean score, and expected mean were calculated for all plant behavioral examples. The mean was then subtracted from the expected mean to observe which direction each response leaned toward, either pro- or anti-plant intelligence views.
Themes related to respondents’ answers were manually identified in Microsoft Excel due to the limited amount of text (word count) provided in the answers. Co-occurring thematic explanations were counted for each plant behavioral example (Table 4). Thematic categories were then classified and defined by grouping together related themes and repeated words (Appendix Table 6). A particular theme was identified when it was stated by more than three respondents. Statements in which a theme could not be produced or interpreted due to ambiguity, insufficiency in explanatory strength, reduced word count, and/or irrelevance to the subject matter were omitted. In total, twelve co-occurring thematic explanations were compiled based on respondents’ open-ended answers on case studies of plant intelligence.
Seventy percent of respondents reported that they used TK in their field of study less than half of the time, while 30% reported using TK more than half of the time (n = 120). Opinions of plant intelligence were significantly associated with the frequency of using TK. A higher number of respondents who reported using TK more often considered plants intelligent (χ-squared = 4.091, df = 1, p value = 0.0431, Cramer’s V = 0.21, n = 92). Opinions on the value of TK according to plant scientists were significantly associated with professional background, thus separating ethnobiologists (n = 30) from non-ethnobiologists (n = 111) (χ-squared = 8.055, df = 1, p value = 0.005, Cramer’s V = 0.23, n = 141). A larger proportion of respondents (69%) considered that TK has been undervalued by plant scientists, with a larger share of ethnobiologists sharing this opinion. The value of TK was not associated with participant age nor frequency of collaboration outside the biological sciences (thus with the humanities or social sciences). Despite the fact that the majority of respondents did not include TK frequently, 64% agreed that TK could be useful in their field of study and that TK could help close study gaps (53%) (n = 143). Respondents who did not consider TK helpful to close study gaps were more likely to be non-ethnobiologists (47%) (χ-squared = 22.6, df = 1, p value < 0.0001, Cramer’s V = 0.40, n = 143).
More than half of all respondents (58%) were familiar with the term “plant intelligence.” There was no significant association between opinions of plant intelligence and the following variables: (1) professional background (ethnobiologist or non-ethnobiologist sensu stricto) (n = 114); (2) type of profession (professional researcher or graduate student) (n = 144); (3) age (n = 103); or (4) familiarity with the concept of plant intelligence (n = 141). Sixty-nine percent of respondents initially considered plants intelligent, 20% did not, whereas 12% stated that their opinion depended on the definition of plant intelligence (n = 123). When comparing respondents’ first and second opinions of plant intelligence (measured, respectively, directly at the beginning and later in the questionnaire as an aggregate scoring based on a set of plant behavioral examples of plant intelligence), only one in five respondents (21%) changed their opinion, whereas most individuals (79%) maintained the same opinion.
The results of the plant behavioral examples had an internal consistency of 0.96 (Cronbach’s alpha) which demonstrated very good reliability of respondents’ answers (Table 1). Sixty-four percent of respondents scored the plant behavioral examples homogeneously (i.e., they responded with the same opinion) (n = 114). Ethnobiologists did not differ significantly from non-ethnobiologists in terms of consistency of their answers. Within the latter group, the examples of mere behaviors had more counts of “evolutionary processes” as explanation than the examples of intelligent behavior (Table 4). Furthermore, only two respondents were able to discriminate between true and false plant intelligence behavioral examples. Most behavioral examples of plant intelligence had a higher Likert mean score than the expected mean (2.5 for examples 3–9 and 3.5 for examples 1 and 2), indicating that the direction of the responses leaned slightly toward pro-plant intelligence views (Table 1). However, habituation in Mimosa pudica L. was the only exception with a mean Likert score of 3.4. Compared to conditioned learning and other flexible plant behaviors, a habitual response may have seemed less impressive to participants.
The two examples of mere behaviors had a lower Likert mean score than expected (2.5), indicating that the direction of responses leaned toward anti-plant intelligence views.Table 1A list of the plant behavioral examples provided in the questionnaire where respondents were asked to rank on a scale from one to four (or six) whether they considered the presented behavior to be intelligent or not (1 = not intelligent to 4 or 6 = very intelligent)Plant behavioral examplesa% distributionnMeanb123456(1) Habituation in Mimosa pudica L Habituation is a decreased response of an individual to a stimulus that is being presented repeatedly (your response to a new sound diminishes as you become accustomed to it; traffic, for instance). Can plants become accustomed to a repeated stimulus? Mimosa pudica L. (the sensitive plant) is well known for quickly folding its leaves in response to touch. It is possible that Mimosa exemplars stop folding their leaves after being exposed to frequently repeated stimuli. Gagliano et al. [15] studied habituation in Mimosa by repeatedly dropping potted individuals from a certain height. As they reported, the plants in effect stopped folding after a number of drops. The experiment discarded the alternative hypothesis that the leaves had stopped folding due, for instance, to motor fatigue after the continuous drops, and not to habituation. Intriguingly, this learned response persisted, lasting up to a month [15]21.114.610.621.119.5131233.4(2) Conditioned learning in Pisum sativum L In Pavlov’s classical conditioned learning, Pavlov was able to show that a trained dog was able to salivate with just a neutral cue (a bell as the conditioned stimulus) in the absence of its food (the unconditioned stimulus). The same type of conditioned learning was conducted in green pea plants, Pisum sativum L. [16].