An information theoretic conception of narrative

Narratives are so pervasive in society, and in the psyches of groups and individuals, that it is impossible to ignore their relevance and impact. However, finding a coherent and serviceable definition for narratives for many applications in the social sciences has been elusive.

This elusiveness of a sound and serviceable definition is perhaps because the term 'narrative', like the term 'information', is very polysemantic, or polysemous. Like many frequently used words in English, it has many meanings. Often, the more commonly used a term is, the more polysemous, or polysemantic, it becomes. This fact alone has often served to limit the use of the concept of narrative in social science and psychology.

Limitation here does not entail neglect. The contemporary arsenal of psychotherapies now includes narrative therapies, and several cognitive scientists who have philosophical dispositions are interested in narrative conceptions of personal identity and the self (which is indeed a large subdiscipline in analytic philosophy and cognitive science). Social scientists have been attempting coherent conceptions of narratives for several decades. There is a relative general stability to the concept of narrative across these disciplines and applications. Roughly, a remembered record of a sequence of events and situations made coherent and bound together by actor-experiencers and causal links.

Still, for many applications in psychology and social science, the available concepts of narrative are often taken to be too abstract, too ill defined, too polysemous, or too vague to be useful in more rigorous and scientific fields. Analytic philosopher Bence Nanay indicates the vagueness associated with one important kind of narrative:

The … is to generalize a pair of concepts … widely used in the history of science, in art history and in historical linguistics – the concept of internal and external history – and to replace the often very vague talk of ‘historical narratives’ with this conceptual framework of internal versus external history. (Nanay, 2017)

Political scientist John Boswell signals similar difficulties in applying concepts of narratives in political science for deliberative democracy:

Concurrent with the ‘deliberative turn’ in democratic theory, research and practice, something of a ‘narrative turn’ has also been occurring in the broader social sciences … There has been a growing interest in the impact of narratives on the way people interact and how they understand the world across a range of disciplines, including philosophy…psychology…organisational studies …and politics and policy…As with deliberative democracy, growing interest has made narrative a slippery concept, one to which individuals bring different baggage. For … this article, narrative is defined as a chronological account that helps actors to make sense of and argue about a political issue (Boswell, 2013)

Boswell notes that not only is there a mismatch between the prevalence of narratives in politics and society and the (under)use of narratives in deliberative democratic theory, but that much political theory eschews mention of narratives. However “narrative should be seen as an especially important aspect of [political] deliberation” (Benford, 2002; Boswell, 2013; Coëgnarts & Kravanja, 2015).

In an investigation of social control narratives, Robert Benford classifies kinds of narratives according to actor type (group versus individual), identifying “two generic types of narratives discernable within social movements: participant narratives and movement narratives” (Benford, 2002). There are clearly broad and consistent efforts to present systematic conceptions of narrative – at least relative to the scholarly task at hand.

There’s little question of the pervasiveness of narrative in contemporary culture and politics. Nor is there much doubt that the reason for this is that narratives are readily consumed by individuals and internally encoded into mental representations.

There remains the problem, however, of achieving a good scientifically serviceable definition of narrative that can also be deployed in the humanities, even given the dubiety that some scholars in the humanities entertain about reductionist, analytic scientific epistemology. Yet, scientism versus anti-scientism (and rejection of the binary opposition between the two) aside: a good scientifically sound conception of narrative is arguably desirable even in the humanities.

In addition to the abovementioned ambiguity, polysemy, and vagueness of the concept, there is the problem of how narratives are represented in the brain-minds (a hyphenation used by cognitive scientists to refer to the mind as sustained by the brain) of individuals. One reason for this is that narratives are not necessarily only linguistic. They are also conveyed and sustained by visual and aural information that is not textual in origin. Moreover, they have abstract content in at least the sense of abstract that means an abstraction away from, or out of, the information of a situation or situations.

As I am not an ontological Platonist, I doubt that abstract narratives really exist in a traditional Platonist sense of being abstract. (This is, however, certainly a way in which they are used by many scholars in the humanities). I favour a conception that regards narratives as a kind of universal based upon sets of particulars. The philosophical argument between Platonists and nominalists about abstract entities is not a hill to die on in this post. That being said, it is the nature of the sets of particulars – information sources and messages that comprise narratives - that is largely at issue, in my view.

Another problem is the longstanding issue of how the brain-mind stores language in the first place. Therein lies a hint about how to approach the ontic basis of narratives using information theory. Perception and encoding of textual information is known to be largely phonological, and based on the function of what psychologists call the phonological loop, which has evolved to aurally process proximal stimuli (to process spoken language and other sounds.) One qualitative type of information is encoded using another type, and the types in this case map fairly neatly to what philosophers call natural kinds (sound signals versus visual or photonic signals and stimuli). This certainly coheres with another definition of narrative due to Bence Nanay:

A narrative is able to make intelligible our experiences and feelings…it is more than a way of classifying texts: narrative is a perceptual activity that organizes data into a special pattern which represents and explains experience. (Nanay, 2009)

Using information theory to define narratives

The visual-cum-textual-cum-linguistic nature of narratives, and their multi-modal representation in the brain-mind, mean that the sets of particulars (events, situations, entities, experiences) that comprise a narrative are qualitatively, and ontologically, type heterogeneous. An actual elephant is not the same thing as a picture of an elephant, a dream about an elephant, or a description of an elephant. These are different types of encoding of different types of information. (Measures of information are quantitative, and do not need to be considered as heterogeneous. However, there is no reason to prohibit applying a conventional label to what ontic type of information is being measured, and tagging the quantity accordingly. This is something further explored in my monograph and in (Long, B. An Information Theoretic and Memetic Conception of Narrative Representations, 2020 in preparation.))

An aetiology of narratives – a theory of how they are caused and of their causal inputs – should capture and account for this informationally heterogeneous nature. It should account for the fact that narratives are caused by, and based upon, many different types of information source and message.

Contemporary political narratives provide one of many good examples of this information source type-heterogeneity. Political narratives are arguably nowadays most often transmitted through the medium of The Internet (Admittedly, I am yet to investigate any hard statistical data about audience size of competing mediums, but if broadcast television is still the mainstay, my argument is not disturbed). This heterogeneity of information types and inputs lends a clue about why, apart from the abovementioned problems with vagueness and polysemantic nature of the concepts of narrative, there has been limited use of the concept of narrative in much of social and political science. The multifarious narratives that they call upon and reference, and sometimes represent, are likewise presented multi-modally.

The Internet and World Wide Web are now largely visual, yet representationally and semantically multimodal, mediums for the transmission and exchange of information. The basis of The Web – the technology of hypertext and its underlying hypertext transfer protocol – ensures that text and images are comprehensively combined. The heavily interactive nature of Web media also increases the interlocutory nature of much of the semantic information exchanged.

The multidirectional exchange of information facilitated by the additional communication functions of The Internet and Web – the basis of social media – adds a group interlocutory type of high level (at a high level of abstraction) encoding of semantic information to the set of heterogeneous particulars that comprise the information of narratives. The Web is a space of combined visual and textual representations: of natural language encoded texts and high fidelity visual information. Our perceptual and cognitive representations thereof reflect this.

What we must deal with when defining narratives is informational type heterogeneity both inside the head, and outside of it. Nanay has astutely observed that “a general account of narrative needs to be able to cover both pictorial and literary cases…none of the most influential accounts of narrative are capable of this.” (Nanay, 2009). Nanay thinks that “the way we experience narratives may be helpful in understanding the nature of narrative.” An important issue to deal with is the degree to which heterogeneity of external representations of narrative in the world influence the configuration of internal representations of narrative in the head, something which I explore in my forthcoming book.

I favour an approach influenced by information theory that regards narratives as much as a heterogeneous set of sources and messages as a chronologically ordered set of semantic messages or descriptions that might be formulated or encoded from such sources. To bring this out more clearly, there are evidently any ways in which narratives can come into existence and be transmitted. Say you are watching your favourite television show – The Handmaid’s Tale (of course!) The narrative that is presented to you is pre-fabricated. If we are to give Iser’s theory of reader reception and response some credence (I do) then the narrative is also significantly constructed by you at time that you receive it, based upon your own internal ideas and concepts, and your own existing internally represented narratives (which are also classical information sources). The result is a synthesis of semantic information resulting in augmented and partial internal representations of the transmitted narrative (who remembers every detail of every movie scene?)

However, you do not have to receive or consume a narrative in partial pre-fabricated form like this: received from an author and the many creative individuals that contribute inputs to the process of making movies and series on The Internet. You may end up unfortunate enough to live in a society not unlike that depicted in the Handmaid’s tale. In this case your unfortunate brain-mind will assemble your own narratives based upon your immediate experience of many information sources in real world situations. Your internal narratives will take on far more of a personal documentary aspect and content. They will involve different information types. They will encode more veridical messages based upon information, rather than pseudo information (fiction).

A notable break from theorising about sequential event and experience based narratives came in the form of the discursive turn in continental philosophy and literary theory. In the mid to late 20th Century discourse became the conceptual tool par excellence for investigating the meaning of literary and cultural texts, and socially shared meanings.

Discourse stood apart from narrative as less (or differently) structured, more heterogeneous, more amorphous, and less sequential. It fitted well with the outlook of post-structuralists and postmodernists who eschewed the idea of stories with beginnings, middles, and ends, and instead embraced concepts like the experiential thrown-in-ness of the German Existentialist Heidegger and the French Existentialist Jean Paul Sartre. There were other motives for the move to discourse, including the effort to deal with and respond to what Foucault and Francois Lyotard called grand narratives and metanarratives. Especially those of religions.

It would be remiss not to call upon the philosophy of religion with respect to narratives, even when information theoretic conceptions of narrative and their grounding in applied scientific and applied mathematical theories of information and communication is my primary objective. I suspect it is plain to most readers that most religious traditions trade in narratives as a primary semantic and moral currency. As such, it is likely no accident that theist philosophers like Paul Ricouer, Rene Girard, and Alasdair MacIntyre have placed a great deal of emphasis on narrative and storytelling (and, in Girard's case, authors) in their social theories and philosophies (Girard, 2017; Ricoeur, 1991, 1998). There is a foundational theological ontic component to, and motive for, this habit. The God character of the Christian Bible spoke the world into existence in the Genesis creation narrative.

The entire corpus and tradition of Christian theology is thoroughly narrational, interlocutory, and foundationally linguistic and textual in nature. Our oldest religious traditions, unsurprisingly, rely heavily upon one of our oldest and most important human technologies: language. There would arguably be limited hope of their ever having been transmitted and sustained otherwise. (Iconography and spiritual imagery seem largely supplemental to texts in most theist traditions. I cannot speak to the idea that persons of faith claim to somehow receive both mental and visual religious images directly, and my only point of reference in this connexion is psychological science and the study of hallucination).

Like discourses, abstract (cultural) texts, tropes, and meanings, messages are linked with, and intrinsic to, narratives. According to my approach they are 'half' of their ontic basis, with the other 'half' provided by information sources (it is a useful principle of classical information theory that representations can also be modelled as sources, and as messages).

Necessarily, messages in the context of narrative are meaningful. They have semantic content. Messages in classical applied scientific information theory, however, are stripped of their meaning (there are accounts that disagree, but they are out of my scope here) (Adriaans, 2010). The founders of classical mathematical communication theory – Claude E Shannon and R.V.L Hartley – famously asserted that the engineering question of how to get sequences of symbols from one point to another was separate from the meaning or psychological aspects of those sequences.

An easy way to see why this works is to realise that the same meaning might be conveyed in two different languages. The symbols will differ, but the conveyed meaning will be the same. Another way to understand the separation of semantic content from encoded information bearing messages is to consider homonyms, or else homonymous symbols, in one language.

According to my approach, narratives and discourses are both grounded by, and reducible to, sets of semantically imbued information sources and messages. In a sense this approach stays true to discursive theorists in that it allows narratives to be regarded as a type of discourse structured in a certain restricted way. That outcome, however, is incidental to my more analytic and scientifically attuned purposes.

Just as meaning and syntax is separated in atomic language and logic, information sources and messages can be separated from those meanings to which they are bound by convention and psychology (there are other ways in which they can be so bound). However, the matter is not quite so simple, and this cleaving itself is largely conventional. Moreover, the basis of the semantic content of information is not, by any means, a matter agreed upon. Several serious attempts at producing theories of semantic information have not succeeded yet in a broadly ratifiable theory (Allo, 2010; Carnap & Bar-Hillel, 1952; Dretske, 1981; Floridi, 2004). This is one reason for the pluralism about the nature of information and semantic information that dominates the field.

With some caveats associated with the type of information one is dealing with (natural physical information versus cognitively encoded artefactual information, for example), information is best thought of as what some analytic philosophers call a truth maker, rather than as a truth bearer. This approach is fairly alternative, as the truthbearer conception is probably currently more popular (I say probably, because much philosophy and theory of information and the nature of information is subject to an overarching definitional and conceptual pluralism (Bueno, 2010; Shea, 2013.)) It requires what is known as the veridicality thesis: that (semantic) information must be alethic (truth apt, or able to be true or false) and true in a similar way to a true statement or assertion is true.

Theories of truth, or of how statements, concepts, and representations are made true or false, are a big deal in philosophy and formal logic. I will not venture into that well-trodden and hotly contested territory here. However, it is important that some theorists do not even consider a separate theory of semantic information necessary, believing the resources of indication and classical statistical theories of information to have enough resources to sustain a basis for semantic content (Adriaans, 2010). Moreover, some pluralists require truth-aptness for semantic information, but leave the question of the meaning of natural low-level physical information ambiguous.

Leading philosopher of information, Luciano Floridi, presents what is probably the most mature of the truthbearer conceptions of semantic (meaningful) information. According to Floridi’s approach, (semantic) information must encapsulate truth. However, there are ways in which much more basic kinds of natural and physical information can be thought of as having meaning, but without being bound to the concept of truth (Adriaans, 2010; Long, 2014).

Causal indication is one such approach that has its origins in research about the meaning of representations such as symbolic descriptions and depictions. One of its advantages is that it coheres with, and can sustain, proportional transmission of information, thus avoiding the need for messages to be alethic on a wholly true-false (alethic) basis. However, as useful and indispensable as it is for the purpose, a heterogeneous and hybrid approach to semantic content is necessary for fulfilling the functional and actualising role of meaning making for complex heterogeneous sets of information sources and messages.

Narratives (including symbolically and syntactically based representations) are reducible to, and thus definable as a set of particular sources and messages, and their semantic contents or meanings. Narratives can also be representations of such (albeit narratives of a different type). The semantic content of such narratives is based upon the semantic information of the sets of sources and messages. The semantic content of the sources and messages involved cannot be established simply according to any given theory of truth, and involves a pluralist combination, or hybrid, of correspondence theory, representation by causal indication, semiotics, association, and, in some cases, modal semantics (Long, B. An Information Theoretic and Memetic Conception of Narrative Representations, 2020 in preparation.) This pluralism about semantic information is driven by the heterogeneity of semantic source types, but also simply by the fact that no theory of semantic information that defers to only one approach appears, so far, to be able to satisfactorily account for semantic information.

References

Adriaans, P. (2010). A Critical Analysis of Floridi’s Theory of Semantic Information. Knowledge, Technology & Policy. https://doi.org/10.1007/s12130-010-9097-5

Allo, P. (2010). A Classical Prejudice? Knowledge, Technology & Policy. https://doi.org/10.1007/s12130-010-9098-4

Benford, R. D. (2002). Controlling Narratives and Narratives as Control within Social Movements. In Stories of Change: Narrative and Social Movements.

Boswell, J. (2013). Why and How Narrative Matters in Deliberative Systems. Political Studies. https://doi.org/10.1111/j.1467-9248.2012.00987.x

Bueno, O. (2010). Structuralism and information. Metaphilosophy. https://doi.org/10.1111/j.1467-9973.2010.01641.x

Carnap, R., & Bar-Hillel, Y. (1952). An Outline of a Theory of Semantic Information (Issue 247).

Coëgnarts, M., & Kravanja, P. (2015). Embodied cognition and cinema. Leuven University Press.

Dretske, F. (1981). Knowledge and the flow of information. Blackwell.

Floridi, L. (2004). Outline of a Theory of Strongly Semantic Information. Minds and Machines. https://doi.org/10.1023/B:MIND.0000021684.50925.c9

Girard, R. (2017). Evolution and conversion dialogues on the origins of culture (Bloomsbury). Bloomsbury Academic.

Long, B. R. (2014). Information is intrinsically semantic but alethically neutral. Synthese. https://doi.org/10.1007/s11229-014-0457-7

Nanay, B. (2009). Narrative pictures. Journal of Aesthetics and Art Criticism. https://doi.org/10.1111/j.1540-6245.2008.01340.x

Nanay, B. (2017). Internal history versus external history. In Philosophy. https://doi.org/10.1017/S0031819117000067

Ricoeur, P. (1991). Narrative Identity. Philosophy Today, 35(1), 73–81. https://doi.org/10.5840/philtoday199135136

Ricoeur, P. (1998). Figuring the Sacred: Religion, Narrative, and Imagination. Pro Ecclesia: A Journal of Catholic and Evangelical Theology. https://doi.org/10.1177/106385129800700314

Shea, N. (2013). Naturalising representational content. In Philosophy Compass. https://doi.org/10.1111/phc3.12033