Axon: Creative Explorations

Introduction

Ours is an era of environmental crises, in which ‘extensive, cumulative and far-reaching effects of human impact on environment’ including climate change, pollution and resource depletion now reach extremes at which ‘the future of humanity is threatened’ (Matthews 2020: 39). Climate change activist Greta Thunberg calls for public dialogues to incorporate diverse perspectives and problem-solving approaches, including those of ‘people who think outside the box and aren’t like everyone else’ (in Jagannathan 2019: n.p.). For Thunberg, who identifies as being on the autism spectrum, this especially means heeding neurodiverse people’s contributions, in line with pushes to conceive neurodiversity as ‘neurological difference, not pathology’ (Savarese & Zunshine 2014: 18). The terms ‘neurodiversity’ and ‘neurodiverse’ pertain to people including but exceeding those on the autism spectrum: neurodiverse people are neurologically predisposed to think and act in ways that exceed the socially-privileged norms of ‘neurotypical’ behaviour (Muzikar 2018: n.p.).

Agreeing with Thunberg and the neurodiversity movement’s aims, but recognising that reasonings that exceed mainstream norms are not the sole prerogative of neurodivergent people, this article takes as a founding premise that dialogues around sustainability should encompass diversities of both neurodivergent and neurotypical ‘mind styles’. Mind styles entail differing problem-solving approaches that offer valuably diverse possibilities (Zhang 2013: 4); their value in scientific teams is illustrated by Devezer et al. (2018: n.p.). Connections between diverse mind styles and environmental sustainability were raised decades ago by Félix Guattari who in his writings on ‘ecosophy’ (ecological philosophy) argued the need for a diverse ‘mental ecology’ (1989: 36—37). Guattari’s approach, though in tension with mainstream evolutionary literary criticism, can be considered in terms of what Professor Liane Gabora terms ‘Self-Other Reorganization’ (SOR) frameworks, which emphasise creativity, collaboration, networking and communal exchange or horizontal transmission as key to collective survival for interdependent forms of life (2019: 7—8). Such frameworks explicitly differ from Darwinian models emphasising linear trajectories of competitive selection — an approach that concerns us and from which we distance ourselves, given its ethically troublesome applications in eugenics and elsewhere (Barta 2005: 116; Ayala 2014: 24).

A challenge for diversifying the mental ecology is that differing mind styles frequently entail differing communication strategies (Zhang 2013: 185). This is the key problem this chapter broaches. We write across disciplinary divides as researchers whose interests lie in cognitive science (Steven Langsford) and creative writing (Amelia Walker). In the first of two main sections, Amelia poses that ecopoetry can increase openness to mind styles that differ from one’s own. This prompts discussion of Guattari’s ‘three ecologies’ (1989: 17), then inquiry into ecopoetry’s attitude and aesthetics. Poems from Australian open-access journals Plumwood Mountain and Cordite help illustrate how ecopoetry prompts openness to new ideas. Section two opens with an ecopoem penned by Steven in response to ecopoetic strategies outlined by Amelia. Steven then provides a cognitive perspective on ecopoetry and the mental ecology. This includes consideration of Bayesian inference (Jaynes 2003: 112), rational speech act (RSA) theory (Goodman & Frank 2016: 818) and communication modelling (Devezer et al. 2018: n.p.; Pöyhönen 2017: 4519).

 

Proposing ecopoetry’s benefits for sustaining diversity in the mental ecology and broadening the imaginable approaches to environmental crises of our times

Amelia Walker: Discussions of ecopoetry emphasise how it supports environmental sustainability through themes regarding ‘the egregious impact of humans on the world and other species’ (Chisholm 2014: 120). I’d like to pose that ecopoetry can additionally promote diversity in the mental ecology.

Steven Langsford: Hold up, this is new to me. What do you mean by ‘the mental ecology’?

Amelia: Felix Guattari’s The Three Ecologies describes interconnecting ‘mental’, ‘social’ and ‘environmental’ ecologies, presenting all three as threatened by monoculturalism — the environmental through species extinction and selective farming practices leading to reduced biodiversity; the social through cultural colonisation of non-dominant social groups by dominant groups and the resulting suppression or erasure of non-dominant cultures; and the mental via elevation of dominant over non-dominant mind styles (1989: 32—37). These ecologies are interdependent: change in one affects the others; cultivating mental and social diversity supports environmental sustainability, for instance by providing more perspectives on ways to address environmental crises (35—36).

Steven: The ‘mental ecology’ concept fits well with the Dawkins-style metaphor of memes of culture and thought as evolving organisms (Dawkins 2016: 24—58), which necessarily requires some source of variation.

Amelia: Yes. Except that while Dawkins focused on ‘selfish’ ways genes and memes compete for survival, Guattari’s approach was closer to that of earlier theorists who focused on ‘mutual aid’ — the idea that survival is most likely through cooperation, sharing and reciprocity (Borello 2010: 33—36). More recently, Gabora has posed the concept of ‘Self-Other Reorganization’ (SOR) frameworks that emphasise creativity, collaboration, networking and communal exchange (2019: 7—8).

Steven: There’s a rich literature in the philosophy of science exploring the benefits of epistemic diversity which resonates with this, for example Devezer et al.’s description of the benefits of strategy diversity in scientific communities (2018: n.p.).

Amelia: Collaboration in scientific teams with different strengths beautifully illustrates a diverse mental ecology’s benefits. A challenge is that those who think in different ways often communicate differently. That’s why I’m suggesting ecopoetry as something that can increase openness to differing mind styles and communication practices.

Steven: How would ecopoetry do that?

Amelia: Discussing ‘poetry and knowing’, Jen Webb reminds readers that language is connected with thought (2009: 5—7). Webb is among multiple theorists who pose that linguistic innovation through poetry expands lexical and grammatical possibilities, which in turn extends scope for generating and articulating new ideas (Gibbons 2015: 7—12; Perelman 1996: 10—14). For instance, Reginald Gibbons poses rhyme as a ‘poetic technology’ that can prompt reflection on connections between words and concepts one might not otherwise think to associate (2015: 58-59).

Steven: Your examples suggest that multiple kinds of poetry might benefit the mental ecology. Why the focus on ecopoetry?

Amelia: There’s two reasons: one, the ecopoetic attitude, and two, aesthetics. Comparing ecopoetry with environmental poetry illustrates the attitude. While ecopoetry and environmental poetry are both ‘directly engaged with politicised environmentalism’, environmental poetry typically positions humans as subjects and nature as a range of objects (Arigo 2008: n.p.). Ecopoetry conversely treats humans as animals — we are part of nature, and should humbly recognise beyond-human entities as bearing modes of wisdom that may sometimes exceed ours (Hume 2012: 751). Virginie Greene makes a similar point about humans as ‘literate animals’, observing that ‘all literates are humans, all humans are animals, therefore all literates are animals’ (2014: 35). Ecopoetry destabilies human-animal dichotomies through encounters via which readers and writers open themselves to the unfamiliar (Hume 2012: 762).

Steven: Can you give examples?

Amelia: In ‘Two Meditations on Ecology’, published in open access literary journal Cordite, John Hoppenthaler cites the intelligence of whales, remarking ‘How like us they are, beached and prostrate’ (2020: n.p.). Within the same ‘Earth’-themed issue of Cordite, poems by Corey Hill, Jena Woodhouse and Lucy Alexander stage encounters with squids, rainforest snails and crows. Although ecopoetic encounters are typically between the human and beyond-human, I think this encourages openness to the unknown generally, including mind styles differing from our own. These thoughts are informed by notions of ‘neurocosmopolitanism’ (Savarese & Zunshine 2014: 17) and work on ‘cognitive alterities’ that forges connections between art, literature and the traversing of self-Other boundaries (Chance 2012: 247).

Steven: That covers the attitude. What about aesthetics? Are there particular literary techniques associated with ecopoems?

Amelia: Ecopoetry arises from practices of L=A=N=G=U=A=G=E and visual poetry, among other sources (Chisholm 2014: 120). These histories are reflected through ecopoetic experiments with language and layout (Hume 2012: 760). An example is ‘sraM saM’ by Patricia Sykes in the open-access ecopoetry journal Plumwood Mountain.

Steven: Oh, seems like the title invites a bottom-upward reading that mimics the flight from Earth? Interesting device.

Amelia: Right. The title requires reading in a non-habitual way. Within the poem, Sykes hangs a set of ‘» » »’ symbols over a line evoking the 'flight of the im/possible’ (2020: n.p.). The ‘»’ line is readable as an allusion to birds, planes, spaceships and more. Later, there’s a display of four capital ‘O’s that collectively create the idea of a bigger ‘O’. This could be a mouth or a black hole or the space left after an explosion … Or something else entirely. The ambiguity is key. Each reader has broad scope for constructing their own meanings.

Steven: Sykes also seems to play with space. For example, some lines and stanzas are aligned completely to the left. Others are set at a hanging margin. This creates a lot of typeless space.

Amelia: Yes. This is another common ecopoetic technique. In the same Plumwood Mountain issue it features in poems by Shari Kocher, Frances Presley and Jennifer Mackenzie. Ecopoet and critic John Kinsella deploys it liberally throughout his oeuvre. In his critical writings, Kinsella explains how typeless spaces in ecopoems provides visual metaphors for the emptiness left behind by environmentally destructive acts of land clearing, resource consumption, hunting species to extinction, and more (Kinsella 2013: 11).

Steven: Yet Heather Taylor-Johnson’s and Alana Kelsall’s poems in that same issue of Plumwood Mountain seem to do the complete opposite. They’re basically prose poems.

Amelia: They are. Reading them with knowledge of how ecopoets use typeless space lends special meaning to this. These overcrowded poems metaphorically suggest human tendencies to colonise, crowd out and overrun spaces. They also signal excesses of production and consumption.

Steven: Okay. Now we’ve clarified key features of ecopoetry in the form of attitude and aesthetics. How might one go about writing an ecopoem?

Amelia: In a recent writing workshop, I encouraged participants to refer to a beyond-human o/Other in second person (as ‘you’) and themselves in third person (as ‘she’, ‘he’ or ‘they’) — a grammatical shift in perspective that encourages a shifted relationship to self and subjectivity.

Steven: I’m not totally convinced this is possible yet, but it sounds like a great thing to try.

 

Poetic language and epistemic diversity: a vital relationship

‘Squirrel’
Poem by Steven Langsford

Like deer, they do not respect kingdoms,
Drifting across your territory like clouds.
They are the bin-fillers, heavy and tall,
The chattering hand-wavers, the tailless.

Although they cast cloud-shadows not hawk-shadows,
It is easy to fear them.
With their bark-crusted humps, rubber hooves
And flashing eyes
Wires run from their ears to their stomachs,
Metal grows from their fingers and necks.
They are loud.

It is not their attention that is dangerous, but their blindness.

Their seasons do not follow the tree’s seasons.
Their paths run only on the ground, from bin to bin.
Paths that run through your kingdom.
But sometimes
They will realise where they are
And kneel, eyes flashing, howling softly,
In their outstretched hand an offering,
Yours to judge. If this is acceptable.

 

Steven: Would this fit as a response to the writing exercise you described? The ‘they’ in this poem represents humans and the ‘you’ (‘your kingdom’; ‘Yours to judge’) is the squirrels.

Amelia: It’s perfect. I love how the ending flips anthropocentric assumptions that humans name and judge beyond-human entities, instead inviting readers to wonder how squirrels might evaluate our actions. The wires running from ears to stomachs spark my curiosity. I’m guessing this refers to headphones for music or devices …

Steven: Well, they could be —

Amelia: Don’t tell me. The uncertainty makes me push beyond the limits of what I already know. That’s how ecopoetry opens minds. I’m curious to know how it felt to write it. Did the process reveal anything new to you?

Steven: The third and second-person combination and your description of why you wanted that constraint were what grabbed me about this exercise. It gives a very detached and alien voice that isn’t from any of the perspectives in the poem, human or animal. The outsider’s view of the humans should be the squirrel’s, but addressing the squirrel as ‘you’ forbids that interpretation very early on. That freed me from having to actually take on the squirrel’s perspective, which I don’t think I could do (and it wouldn’t look anything like this).

Amelia: How would the idea of ecopoetry promoting diversity in the mental ecology be perceived through a cognitive sciences lens?

Steven: Cognitive modelling is a long way from catching up to poetry. But it has some things to say about communication that might be interesting to you. In particular, this description of the relationship between the mechanics of how ecopoetry works and neurodiversity echoes an argument made by Goodman and Frank (2016: 818—29) in their rational speech act (RSA) account of how communication might be well-described by Bayesian inference (Jaynes 2003: 112).

Amelia: I’ve not come across RSA before …

Steven: It’s an explicit mathematical model sketching a way to ground meaning in pragmatics. The two main assumptions are the mathematics of sound probabilistic inference and the idea that ‘language comprehension in context arises via a process of recursive reasoning about what speakers would have said, given a set of communicative goals’ (Goodman & Frank 2016: 819). These minimalist assumptions imply a lot of rich communicative behaviour. The recursion makes the inference technically difficult, so it’s still under development and it’s an open question how well it describes what people actually do.

Amelia: What’s the link?

Steven: The reason I’m bringing this up is for the contrast between the original RSA and the later uRSA extensions (Goodman & Frank 2016: 818—29). The ‘u’ is for ‘uncertain’ and refers to a version of the model that allows for joint inferences about both the speaker’s intended meaning and other aspects of the interaction. The other aspects include things like uncertainty about the topic of discussion, or uncertainty over how similar the conversation partners are in their background understanding of the world. It turns out including that uncertainty is critical to motivating non-literal language.

Amelia: I’m likewise unfamiliar with Bayesian inference. A basic literature search tells me it’s based on Bayes’ Theorem: a mathematical formula for calculating conditional probabilities (Joyce 2019: n.p.). It’s what you get if you take the probability of an observation given some conditions, and then rearrange it so that it states the probability of the conditions given the observation. One typical example might be medical diagnosis. It’s relatively easy to say what the probability of various symptoms are for a range of diseases. Coughing is high probability in both colds and lung cancers, fever is high probability for influenza but low for cancers, and so on. Only none of these probabilities are the ones patients and doctors are interested in. They want to know the probability of each condition given an observation of some symptoms. That requires inverting the conditional probability, using Bayes’ rule to appropriately combine the knowledge about which symptoms are likely with each condition, the observations, and the base-rate incidence of each disease.

Steven: That’s right. The only critical ingredient in being Bayesian is a commitment to representing uncertainty using probabilities (Jaynes 2003: 112). That leap has been controversial in statistics: the frequentist alternative is to use probability to represent some proportion of a reference class. The two look similar, and share terminologies, but answer fundamentally different questions. By appealing to a reference class, frequentist statistics can give guarantees about error rates and about the relationship between your particular random sample and the full population it was drawn from. By appealing to some set of prior beliefs, Bayesian statistics describe the degree of to which you ‘should’ believe some uncertain proposition given evidence about it. Even though both of them talk about the ‘probability’ of an event, they take ‘probability’ to mean qualitatively different things.

Amelia: Another example of how slippery language can be.

Steven: Yes, which Goodman and Frank recognise when they note how ‘understanding language is more than use of fixed conventions and more than decoding combinatorial structure’: comprehension also requires ‘exquisitely sensitive inferences about what utterances mean given [the comprehender’s] knowledge of the speaker’ (Goodman & Frank 2016: 818).

Amelia: Ah yes … that relates to interpreting poetry. But how do mathematical calculations fit in? The conditional factors of communication aren’t easy to look up and plug into an equation like mortality rates.

Steven: That’s part of the problem Goodman and Frank address. The claim that ‘listeners think that speakers have a mental model of the listener, and they choose their words based on how they think the listener will react’ seems ridiculously commonsense, but it’s surprisingly hard to implement. One reason is that full Bayesian inference is computationally intractable, so any theory that appeals to Bayesian principles necessarily relies on some kind of approximation (van Rooij et al. 2019: 182—84). In this particular case, there’s a recursive loop where the speaker has a model of the listener who has a model of the speaker who has a model of the listener and so on to infinity. To capture what people are doing you definitely need more than zero levels of recursion, but going to infinity breaks everything.

Amelia: How do they address these challenges?

Steven: Partly by simplifying the environment, so that instead of dealing with all of language they deal with communication in simplified games. The earliest were reference games, ‘guess which object I’m trying to indicate from this set’. The set of scenarios treated by RSA keeps expanding though, and now include things like politeness in requests and hyperbole. Although even the most recent versions of RSA treat very limited scenarios, I think there’s a lot of value in studying patterns in the things they can and can’t do. That’s the kind of evidence I’m appealing to in this argument that there is a fundamental connection between diversity and creative non-literal language use. Early versions of RSA were implemented with speakers and listeners that were identical clones of each other. This simplifies the communication problem: to infer what the speaker meant by some message, the listener can imagine what would prompt them to use the same message. Many features of language thought to be distinctive of cooperative communication (Grice 1989: 36—41) can be reproduced in this simple model, but non-literal language is not one of them.

Amelia: Can you give examples?

Steven: The Gricean maxim of ‘quantity’, the idea that cooperative communicators should prefer to be concise, can be captured by adding a cost that rises with message length (Bergen et al. 2012: n.p.; Degen et al. 2013: n.p.). Other tweaks to the utility function that speakers seek to maximize can capture things like indirect language motivated by politeness, or language intended to achieve a social purpose such as insulting, comforting, or flattering the listener rather than just drawing attention to particular states of the world (Yoon et al. 2016: 2771). But none of these adaptations give those early RSA agents any reason to use hyperbole or figurative language.

Amelia: Aha … That creates challenges for poetry.

Steven: Yes! The key change to the model that motivates the use of non-literal language is adding uncertainty about the other partner in the interaction as well as the state of the world (Goodman & Frank 2016: 818; Noveck 2018: 169—71). When the listener is uncertain about how closely their worldview matches the speaker’s, sound inference requires them to use their background knowledge about how the world usually works, which ‘dampens’ the effect of the message if the speaker stays very literal.

Amelia: Again, could you offer examples?

Steven: So if you're at a poetry reading and you want to make a guess at the poet's height following the normative principles of Bayesian reasoning, your guess should be informed by both the evidence of your eyes and your knowledge of the typical distribution of heights. The evidence of your eyes is always uncertain. It can be distorted by things like distance, lighting, posture, or angle. If you tend to trim down your estimates when your eyes tell you the poet is unusually tall and inflate your guess a little when your eyes suggest the poet is unusually short, on average your guesses will tend to be corrected in the right direction. The same logic applies when you’re making inferences about the contents of the poet’s words. If they are describing the height of a tree, your guess about this unseen tree’s height should be biased towards typical tree heights. If they’re describing a depth of anger, your guess should be biased towards typical amounts of rage. The tendency of listeners to assume that things are more likely to be close to normal presents a challenge for a speaker who wants to convince a listener that something is far from normal. The problem does not arise for identical minds that have exactly the same idea of what is typical. In that case both would know what the other meant by a short or tall tree. But poets who are unsure what kinds of trees their listeners are used to and don’t want their description to be dragged back towards the typical by normative inference can be motivated to use non-literal language. While something like ‘reaching the sky’ might be less clear about the exact height of the tree, it could be more effective in reliably supporting the interpretation ‘very tall’ (Kao et al. 2014: 12202—207).

Amelia: That makes sense … Although I haven’t seen hyperbole used very much in ecopoetry, which tends towards under- rather than overstatement.

Steven: That’s a great example of that recursive nature of communication that makes the implementation of RSA so hard. If speakers and listeners both know that literal language gets dampened by normative inference and hyperbole is needed to communicate extremes, then very obvious understatements become a signal that I could have used hyperbole but didn’t: the meaning must be unreachable by both literal and hyperbolic language, it’s that extreme. That’s two levels of recursion on top of the literal meaning. It’s hard to know where to stop.

Amelia: So overall, how do you — from a cognitive sciences viewpoint — perceive ecopoetry’s potentials for promoting communication between diverse mind styles?

Steven: There’s still a gap: while ecopoets are pushing the boundaries of what language can do, quantitative models of pragmatic inference-based communication are getting pretty good at playing ‘Guess who’ (Khani et al. 2018: 543). Despite that, this discussion is a nice example of where even limited models can communicate an important point. In this case, even admitting that RSA (clones) and uRSA (diverse agents) are both very limited, the specific ways in which they differ give an account of why saying nothing or talking in gibberish might be the very best way something can be said.

Amelia: That makes sense to me.

Steven: Another point of contact between cognitive modellers and this ecopoetry discussion might be the ongoing efforts to study the value of diversity in scientific strategies by looking at the behavior of simplified model systems. These models often find advantages for epistemic diversity (Devezer et al. 2018: n.p.; Pöyhönen 2017: 4519). Like RSA, these models don’t predict the outcome of any particular experiment, but they do make explicit a system of relations. In this case linking the modelling assumptions, the kinds of epistemic diversity they admit, and the resulting behaviour. The models are heavily artificial, so there is always some tension between realism and effective simplification (Frey & Šešelja 2018: 407; Evans 2019: n.p.; Navarro 2019: 28).

Amelia: Why express an argument via this format?

Steven: One reason is the way it admits only a very narrow (and constructive) kind of response: if there is some critical element missing it is up to the dissenter to provide it and check the new dynamics; if an assumption is too strong it is up to the dissenter to relax it and show the new dynamics (Guest & Martin 2020: n.p.; Alexander et al. 2015: 242). Another reason for expressing the argument for diversity in terms of a model community of software agents is that the system of relations that is exposed makes no appeal to aesthetics, ethics, or morals. Aesthetic arguments for diversity have no answer against symmetrical aesthetic arguments for homogeneity. But the simplified models of communication and the simplified models of scientific discovery cited here suggest that any personal aesthetic or moral stance has to be made in the context of an independent relationship that exists between epistemic diversity, exploration, and poetic language.

 

Concluding thoughts

Through our dialogue, we have considered challenges of fostering a diverse mental ecology, which requires communication between people with differing mind styles. Ecopoetry bears two features that facilitate this: one, the ecopoetic attitude encourages openness to the unfamiliar, and two, ecopoetry’s formal techniques push readers and writers to think in ways that exceed habitual thought practices. We have connected theories about ecopoetry with cognitive sciences perspectives on communication and thus articulated two distinct approaches to reciprocal interrelationships between epistemic diversity, exploration, and poetic language. There remains more to explore regarding these connections and possibilities. Dialogues that bridge the arts and the sciences seem rich in scope for continuing such explorations.