What is language? The answer one might get to this question is likely to be entirely dependent upon to whom it is asked. If you ask a primary school teacher, the reply might be that it is the sounds and symbols that we use to communicate our thoughts, if you ask Noam Chomskey, you may find that it is a complex coding language of universal grammar and if you ask Benjamin Lee Whorf, you may find that it is the essential abstract structure of thought itself. As ambiguous as it is, at the very least, for many philosophers (myself included), language seems to be inextricably tied to human culture and by definition, the ways with which we describe our world. In this essay, rather than investigate how the language we speak influences our thought, we will examine how the way we think frames our understanding of the phenomenon of language and hopefully gain some insight that would otherwise go unnoticed. Essential to this undertaking is a particular juxtaposition, originally conceived by Fred Cummins in his 2018 book, ‘the ground from which we speak’. Cummins delineates two very different ways of viewing the world, categorized by the two pre-Socratic philosophers who give rise to the earliest forms of each.
The Parmenidean mode of thinking, according to Cummins (2018) represents the scientific or mathematical view of the world and may be described as seated in eternalism and existence. This view represents the mechanistic world view of Newton, Descartes cogito and the cognitive (read: computational) approach to cognitive science. The Heraclitean mode of thinking, in contrast, represents everything as flux, is seated in presentism and becoming and more accurately represents the enactive and embodied approaches to cognitive science. Metamorphosized, these could be viewed as akin to the difference between how some (incorrect) folk theories of mind compare describe different sides of the brain and different people as thinking either more artistically or logically or the juxtaposition between the arts and STEM.
The Parmenidean viewpoint
The traditional approach to cognitive science (and linguistics for that matter) is strictly Parmenidean. The views of Reines & Prinz (2009), Pinker (1994), De Villiers & De Villiers (2003), Fodor (1987) and others, and most of the linguistic relativity debate are all saturated in this mechanistic approach as they attempt to quantise the very structure of thought itself, reducing sentences and their meaning into discreet bits of information regulated by phonemes, grammar and syntax and try to distil from this the relationship between talk and thought. This mechanistic, scientific world view will serve as our starting point. Here the cogito is couched in the representational or computational theory of mind where “cognitive states and processes are constituted by the occurrence, transformation and storage (in the mind/brain) of information-bearing structures (representations) of one kind or another” (Pitt, 2020). Of all of the various theories of consciousness available, integrated information theory (IIT) appears to be the most promising, purporting to be able to tell us something about both the quality and quantity of consciousness in any given system (Oizumi, Albantakis & Tononi, 2014).
Delving solely into only the most relevant parts of the theory itself (as anything further would result in an explosive word count), IIT defines consciousness as integrated information or “the differences that make a difference” (Oizumi, Albantakis & Tononi, 2014) within any given system and provides an algorithm that describes how this mechanistically works. It is an identity theory in which a system of elements (OR, XOR, AND gates, etc) can combine into mechanisms which generate mathematically defined multi-dimensional concept spaces that correspond to past and future states of the system. Of any set of mechanisms in a given system, the one that has the most irreducible information (maximally irreducible conceptual structure, or MICS) is the part that is conscious, with its MICS being identical to its experience and its concept space now being called its qualia space (Oizumi, Albantakis & Tononi, 2014). A concept is defined as “the mechanisms causal role within the system from the intrinsic perspective of the system itself” (Albantakis and Tononi, 2015, p. 5475). The concept that generates the most integrated information (maximally irreducible cause-effect repertoire or MICE) is the one which is conscious in any given experience (i.e. the system of mechanisms that if divided, would result in the greatest loss of information) and is thus called a complex. These concepts can combine into hierarchies where
the maximally irreducible cause-effect repertoire (MICE) of each concept within a MICS specifies what the concept is about (what it contributes to the quality of the experience… while its value of irreducibility QMax specifies how much the concept is present in the experience.
(Oizumi, Albantakis & Tononi, 2014)
This view of consciousness allows for what we may imagine as the conceptual zombies of David Chalmers (1996) in the form of feed-forward computational systems that process information and produce outputs that, while outwardly appearing to behave in a conscious manner, have no “what-it-is-likeness”, thus answering part of Pinkers question about “what kinds of representations and processors the brain has” (Pinker, 1994, pp. 78) and why the Turing machine in fact does not think in mentalese. Although the I.I.T. literature does not as of yet have much to say about language, the recent flurry of hopeful activity from the standard bearers of the Parmenidean worldview, physicists and mathematicians such as Zanardi, Tomka and Venuti (2018) and Kleiner and Tull (2020a,b) as well as the upcoming dedicated issue of the journal entropy seems to throw some weight behind it. By taking it as a given, we may commit ourselves to some speculation.
I.I.T. as an identity theory, correlates potential conscious experiences with their respective concept spaces (a concept space here, is simply a geometric graphical notation that defines past and future states of a system). If a concept (e.g redness) is conscious in an experience, it becomes a complex and its concept space (the particular corresponding states of it’s AND, OR, XOR, etc gates) is then defined as it’s qualia space and is identical to it’s experience.
To elaborate what language may be in this understanding, we may like to imagine a metaphorical Adam and Eve and the very first conversation. Previous to this, communication must have been shaped solely by what was present in the immediate environment and expressed in a concrete manner (e.g. by pointing, eye-gaze and vague vocal eruptions of pain, fear, joy etc). Imagine, if you will, that over time, this primordial couple shared many meals together, often alongside distinctive grunts or vocalizations of satiation. We may imagine these vocalizations as being more akin to a sigh of satisfaction rather than a true language. As the metaphorical couple go about their daily lives, one day Eve, experiencing a feeling of hunger, turns to Adam and emits this sound, the sound that is usually only heard during the process of “eating”. In Adams mind, years of habit immediately make cognizant the experience of (elicit the complex and qualia space identical to) ‘eating’. Because of their shared experience, Adam instinctively knows that Eve must be thinking about ‘eating’, yet it is obvious to him that no food is present. This must be the desired experiential state of Eve rather than a reflection of what is concretely available in the immediate environment. Over successive ‘conversations’ and generations, the abstract vocal sounds that denote specific experiential states (complexes and their identical qualia spaces) become more and more elaborate and articulate until we have a fully-fledged language with a grammar than can code such abstract concepts as the flow of time, ‘if, then’ statements and various abstractions of increasing complexity. New words in this new means of communication can be generated when two speakers come together to share a particular experiential state (complex and identical qualia space), initially in concrete fashions (such as skinning an animal or preparing shelter) and subsequently, through the evolution of dialogue, as the language becomes increasingly abstract and complex, things like abstract mathematics become conceivable. Once two members jointly experience a particular qualia space that has yet to be defined, one may offer up a new type of vocalization and if the two agree, a particular code, or to give a modern video game metaphor, checkpoint, for this particular complex is marked and a new word is born. The internal conscious experience (the complex its identical qualia space) of the speaker can be expressed through a vocalization that immediately makes the same complex cognizant for the listener. As the language develops in complexity it’s recursiveness and compositionality enables speakers and listeners to construct complex complexes and qualia spaces that they may not have direct, first-hand experience of, provided they understand the meaning of the individual words (complexes and their corresponding qualia spaces). Although grammar does not code complexes in the same way as general vocabulary (nouns, verbs, etc), it provides relational stances between them. The more this new language develops, the more this particular cultural group can organise themselves in relation to each other, allowing for more sophisticated division of work and specialization as well as the transmission of learning between both individuals and successive generations, thus gaining an evolutionary advantage over both other non-language using groups of the same or similar species and over their environment as a whole.
This interpretation of the birth of language is no doubt wrong in many ways but the reader should take it as something like a Parmenidean fable of both the teleology (why we have) and ontology (how we have) of language. A way to frame the investigation of language which, it must be said, is intuitive if not somewhat bland and obvious. The important thing to note about this story is that it privileges trivial, habitual and ontological Whorfianism and weak linguistic relativity over linguistic determinism and the language of thought hypothesis (Rescorla, 2019) as language here serves to demarcate qualia spaces that make particular experiential states cognizant rather than as the internal coding system for experience or thought itself, language for thinking rather than language of thinking. To elaborate upon this, we will recall Reines & Prinz (2009) who define the different ways that language effects processing as follows:
Trivial Whorfianism – “when we use words, we draw attention to things we might neglect without it”
Habitual Whorfiansim – languages “install habits of thought that lead us to think in certain ways by default that we would not have thought without language learning”
Ontological Whorfiansim – languages “lead us to organise the world into categories that differ from those we would discover without language”.
Trivial Whorfiansim is expressed as the affordances that are relevant in the environment-language culture dyad are the experiences (complexes and their corresponding qualia spaces) that are jointly experienced as relevant and thus codified in the language, which then cyclically influences the affordances of the speakers. For example, a coastal culture may have more words to describe different aspects of the tide and ocean currents than a nomadic cattle herding culture. Habitual Whorfianism works in much the same manner, as the particular problem solving strategies (shaped by affordances) adopted by the culture are reflected in the language (shared, codified concepts) to the exclusion of other novel strategies that are not developed (or yet to be developed due to a lack of relevance). For example, the Irish language does not have a direct analogue for the English word ‘Yes’ thus an idiosyncratic artefact of a Hiberno-English speakers reply to the question ‘are you going to the shop’ rather than ‘yes’, is often ‘I am’, which has its analogue in the Irish language as ‘Táim’. Something that is not generally heard in British English where the correct answer is simply ‘yes’. Finally, ontological Whorfianism can be expressed again through the affordances relevant to the environment-language culture dyad as the idiosyncratic problems faced by, for instance, a nomadic boreal culture, will differ to those of a sedentary tropical one as the language categories reflect the practicalities of daily living (i.e. a Yupik speaker will see different categories of snow to an Irish speaker, as an Irish speaker will see different categories of rain to a Yupik speaker, due to the environment in which each language developed and is typically spoken).
One point of contention here is the observation that language is necessary to think in certain ways. The most obvious of which is the language of mathematics. If we did not have a word for seventy six, it would be impossible for us to become cognizant of seventy six specifically (as opposed to seventy five or seventy seven). As higher-level arithmetic necessarily requires a language of mathematics, the line between ‘language for thinking’ and ‘language of thinking’ becomes blurred. Ontological Whorfianism may yet come to the rescue however as different numbers and their categories (e.g. natural, real, irrational numbers etc) are once again simply different ways to categorize things mathematically as opposed to ‘more’ or ‘less than’. However, language for thinking may still give way to a language of thinking that is specific to the domain or category of thought (something like a new concept space that allows new affordances in the way that literacy allows us to ‘store’ our thoughts in something analogous to an external hard drive). An evolutionary adaptation that reveals new affordances in a limited variety of Fodor and Pinkers mentalese. Let the linguistic relativity debate rage on, albeit in a rather restricted domain. As the aim of this essay is not to solve this debate once and for all, on this contentious point we shall depart from the Parmenidean frame of reference and move to a worldview in which one can never set foot in the same river twice, that of Heraclitus.
The Heraclitean viewpoint
To address the Heraclitean mode of thought, we will invoke the autopoietic, cognition as sense-making, adaptive model of beings as first put forward by Varella, Rosch & Thopson (1991) and examine how this may map on to the Parmenidean view. Weber and Varella (2002 pp. 115) defined an autopoietic system as
A network of processes of production (synthesis and destruction) of components such that these components:
1. Continuously regenerate and realize the network that produces them, and
2. Constitute the system as a distinguishable unity in the domain in which they exist
(As cited in Di Paolo, 2005 p.5)
Adaptivity is then defined as
A systems capacity… to regulate its states and its relation to the environment with the result that, if the states are sufficiently close to the boundary of viability,
1. Tendencies are distinguished and acted upon depending on whether the states will approach or recede from the boundary and, as a consequence
2. Tendencies of the first kind are moved closer to or transformed into tendencies of the second and so future states are prevented from reaching the boundary with an outward velocity
(Di Paolo, 2005, p.8)
The boundary of viability here means the limit of the systems autopoietic activity, if this boundary is passed, the system dies. Cognition under this view is defined as sense-making which is “behaviour or conduct in relation to environmental significance and valence, which the organism itself enacts or brings forth on the basis of its autonomy” (Thompon & Stapleton, 2009). When two or more entities of this kind come into contact, a new process emerges; participatory sense-making. De Jaegher and Di Paolo (2007, p.8) define this as
the regulated coupling between at least two autonomous agents, where the regulation itself is aimed at aspects of the coupling itself so that it constitutes an emergent autonomous organization in the domain of relational dynamics, without destroying in the process, the autonomy of the agents involved (though the latter’s scope can be augmented or reduced).
For our purposes, this process can be most clearly seen in examples of joint speech (Cummins, 2008 p. 16) defined as “speech produced by two or more people who utter the same thing at the same time”. When this phenomenon occurs, the individual agents synchronize a host of physiological systems such as breath and heart-rate and through the enaction and interaction of the individuals, a new type of joint subjectivity emerges. This type of joint subjectivity can be seen as that of the ninety-nine percent at the wall street protests, the Catholic congregation at a rosary recitation or the respective fan camps at a soccer match. Here language serves not as a message passing procedure but rather as “the enactment of a common world” (Cummins, 2018, p.169) through “the co-ordination of intentional activity in interaction, where-by individual sense-making processes are affected and new domains of social sense-making can be generated that were not available to each individual on her own” (De Jaegher and Di Paolo, 2007, p.13).
Approaching common ground
In order to reach a common ground between our two different viewpoints (and there-by a novel view of language) we will now attempt to correlate some of the processes defined in our I.I.T. based Parmenidean fable with those of autopoietic, adaptive sense-makers. As a starting point, the systems of elements that constitute an individual (its AND, OR, XOR gates etc) would seem to have a direct analogue in the autopoietic system as they constitute the system as a distinguishable unity. Whether or not the system of elements regenerate and realize the network that produces them is slightly less clear although it would appear that this reasoning would provide the only intelligible teleology for such a system to exist outside of intelligent design (which may work for computer models but not so much for living beings); i.e, Eves subjectivity defines her as an individual from her own perspective and generally speaking is concerned with maintaining her own existence in the world.
A systems adaptivity, which regulates a systems states, its relation to the environment and its activity, would seem to be differences that make a difference to the system itself. According to Di Paolo (2005)
Autopoiesis provides a self-distinct physical system that can be the centre of a perspective on the world, and a self-maintained precarious network of processes that generate an either-or normative condition. Adaptivity allows the system to appreciate its encounters with respect to this condition, its own death, in a graded and relational manner while it is still alive.
When Eve needs food to maintain her body, she feels hungry, when Eve feels hungry she searches for and consumes food which then sustains her body and her existence.
We may thus have a system of elements of AND, OR, XOR gates (an autopoietic system) that integrates information to generate differences that make a difference (adaptivity). The enactive model, while not explicitly speaking of experience, defines cognition as sense-making, which is “intentional and expressive” (De Jaegher and Di Paolo, 2007, p.12), when a system “enacts or brings forth a meaningful world” (Cummins, 2018, p. 167). This would appear to have an analogy in the Parmenidean fable in the way that language is generated to demarcate concepts. As “most concepts owe their origin to the presence of regularities in the environment, to which they ultimately must refer” (Oizumi, Albantakis & Tononi, 2014), the concept itself defines the meaning of a mechanism, or, in other words, the meaning ascribed to the systems experience which is directly reflected in the language that any given culture generates to describe its world.
Having outlined some common ground between these two apparently very different ways of thinking, perhaps we should investigate language not as a variety of mentalese (a language of thought), but as a means of participatory sense-making that enables the construction of a shared world through the construction of supervenient ‘subjectivities’ (cultures perhaps?) and common/shared affordances, thus serving as the primary vehicle for the generation of culture.
Albantakis, L., & Tononi, G. (2015). The Intrinsic Cause-Effect Power of Discrete Dynamical Systems—From Elementary Cellular Automata to Adapting Animats. Entropy, 17(12), 5472–5502. doi:10.3390/e17085472
Chalmers, D. J., 1996, The Conscious Mind: In Search of a Fundamental Theory, New York and Oxford: Oxford University Press.
Cummins, F. (2019). The ground from which we speak: Joint speech and the collective subject. Cambridge Scholars Publishing.
De Jaegher, Hanne & Di Paolo, Ezequiel (2007). Participatory sense-making: An enactive approach to social cognition. _Phenomenology and the Cognitive Sciences_ 6 (4):485-507.
De Villiers, Jill G. & de Villiers, Peter A. (2003). Language for thought: Coming to understand false beliefs. In Dedre Getner & Susan Goldin-Meadow (eds.), Language in Mind: Advances in the Study of Language and Thought. MIT Press.
Di Paolo, Ezequiel. (2005). Autopoiesis, Adaptivity, Teleolog Agency. Phenomenol. Cogn. Sci. 4. 429-452. 10.1007/s11097-005-9002-y.
Fodor, Jerry A. (1987). Why there still has to be a language of thought. In Psychosemantics. MIT Press.
Kleiner, J., & Tull, S. (2020a). The mathematical structure of integrated information theory. arXiv preprint arXiv:2002.07655.
Kleiner, J. & Tull, S. (2020b). Integrated information in process theories. arXiv preprint arXiv:2002.07654.
Oizumi, M., Albantakis, L., & Tononi, G. (2014). From the Phenomenology to the Mechanisms of Consciousness: Integrated Information Theory 3.0. PLoS Computational Biology, 10(5), e1003588. doi:10.1371/journal.pcbi.1003588
Pinker S., (1995) Mentalese. In Meier, R. P., & Pinker, S. (Ed.), The Language Instinct: How the Mind Creates Language. Language, (pp. 55-81), The Penguin Press, London doi:10.2307/416234
Pitt, D., (2020 Spring) Mental Representation, The Stanford Encyclopedia of Edward N. Zalta (ed.), https://plato.stanford.edu/archives/spr2020/entries/mental-representation/
Reines, M. F., & Prinz, J. (2009). Reviving Whorf: The Return of Linguistic Relativity. Philosophy Compass, 4(6), 1022–1032. doi:10.1111/j.1747-9991.2009.00260.x
Rescorla, M., (2019 Summer) The language of thought hypothesis, The Stanford Encyclopedia of Edward N. Zalta (ed.), https://plato.stanford.edu/archives/sum2019/entries/language-thought/
Thompson, E., Stapleton, M. Making Sense of Sense-Making: Reflections on Enactive and Extended Mind Theories. Topoi 28, 23–30 (2009). https://doi-org.ucd.idm.oclc.org/10.1007/s11245-008-9043-2
Varela, F. J., Rosch, E., & Thompson, E. (1991). The Embodied Mind. doi:10.7551/mitpress/6730.001.0001
Weber, A., & Varela, F. J. (2002). Life after Kant: Natural purposes and the autopoietic foundations of biological individuality. Phenomenology and the Cognitive Sciences, 1, 97 – 125.
Zanardi, P., Tomka, M., & Venuti, L. C. (2018). Towards Quantum Integrated Information Theory. arXiv preprint arXiv:1806.01421.