Why don’t we read hypertext novels?

Knowledge about human cognitive processing

In much hypertext theory, the open access network structure of hypertext was claimed to mimic the associative nature of the human mind/brain. This idea of hypertext as a ‘more natural’ mode of presentation than the linear and ‘constrained’ mode of print can be traced back to Vannevar Bush’s idea of the Memex (1945), a storage mechanism that would link data and information by way of association rather than hierarchically by indexes. This association-based organizing principle was claimed to mimic the cognitive processing of the human brain. However, Bush’s concept has been refuted by theories of discourse processing (e.g. Dillon, 1996).

Accusing many computer scientists and educators of falling prey to a technocrat dogma, Dillon explicates how such a ‘computer metaphor of the mind’ ¹ fails to capture what is essential about how human information processing occurs:

Naturalistic associationism as manifest in the hypertext literature holds that knowledge is represented cognitively in some form of semantic network or web. The exact form, however, is rarely precisely stated and terms such as schemata and networks, scripts, and webs are employed by writers on the subject with little or no recourse to contemporary psychological developments […]. (Dillon, 1996: 28)

The result, according to Dillon, is an ‘uncritical acceptance of quasi-psychological notions of reading and cognition’ (1996: 27), where claims about the superiority of hypertext over linear print is circulated at the expense of references to established, evidence-based knowledge of how the human mind works. Miall and Dobson (2001: n.p.) go so far as to claim that ‘the embrace of hypertext for literature is possible only for those who have paid little attention to the nature of reading’.

Hypertext theorists’ defective conception of human cognition is also pointed out by theorists and scholars in other fields. Exposing the problems of hypertext design with respect to the ways in which it inherently thwarts human information processing mechanisms, Davida Charney (1994) observes how hypertext imposes cognitive demands exceeding human short-term memory capacity and that arbitrary navigation through a network distorts readers’ ongoing process of establishing meaning and coherence. Miall (1999) suggests that the linked or networked aspect of hypertext not only does not mirror the associative or networked nature of the human mind and its way of responding to literary texts but, rather, that it works contrary to it. Miall states that ‘from the perspective of the reader the inherent tendency of hypertext is, paradoxically, to disconnect text sections, and not to connect them’ (1999: 4).

It may be asked why, if this is the case, hypertext has still been shown to work very well in the case of the WWW. Closer scrutiny reveals that hypertextual presentation on the WWW concerns preponderantly links between unitary texts. That is to say, the WWW does not on the whole offer do-it-yourself paths through unitary texts. Insofar as it does offer intratextual navigation this is of a kind that mimics the same type of navigation available in print texts: for example, between text and footnotes or between chapters or other formal textual divisions that could equally be found on a contents page.

Why and how do we read literature and how does literary reading achieve its purpose

Apart from ignoring basic aspects of cognitive processing in general, hypertext theory has also ignored basic aspects of literary reading. Partly, hypertext theorists’ emphasis on writing was the natural corollary of their ideological emphasis on breaking down the hierarchy between the reader and the writer. But in addition it must be admitted that, compared with the vast number of studies on cognitive aspects of reading (e.g. reading comprehension and metacognition) and the wealth of experimental research using textoids to examine low-level perceptual and linguistic processing, there is a scarcity of empirical research on typical pleasure reading of different genres of literary texts. In this respect, Victor Nell’s Lost in a Book: The Psychology of Reading for Pleasure (1988) has been influential for later theoretical–methodological developments. Although still comprising a small number of scholars, empirical research on literary reading is an emerging, cross-disciplinary field investigating how and why readers take pleasure in reading literature; what psychological, physiological and neurological processes are at play during involvement and engagement in literary texts; and how do these differ from those processes typically at play during the reading of non-literary texts (e.g. expository prose, news)?

Considering the enthusiasm with which its advocates promoted hypertext novels, it is remarkable how little research interest they have triggered outside the electronic literature community. Empirical studies of hypertext novel reading involving actual readers are few and far between, and the studies that have been done tend to find that readers are frustrated, disoriented, confused and tend to get little pleasure from the reading experience (Gee, 2001; Miall, 2004, 2012; Miall and Dobson, 2001). Yet pleasure is exactly what chiefly motivates people to read literature. Norman N Holland (2009) believes that readers come to literature psychologically with two expectations, of which the first is a condition for the second: One, we will not be able to act on it. Two, we expect to take pleasure in the literary work:

Given a literary work, we agree to approach it with Kant’s ‘disinterestedness’. We adopt an ‘aesthetic stance’. We agree just to take pleasure in it. Literature builds on the convention that we will not change the work by our actions. (Holland, 2009: 344)

The inaction described by Holland therefore takes two forms. The first is that the story does not cause us to act in real life, and we inhibit our actions (Holland, 2009; Keysers, 2011). The second follows from the first, that is, we cannot change the work by any action in real life. The pleasure the reader derives from a fictional narrative thus depends on the reader being passive and being ‘transported’ by the reading experience:

when we are ‘lost’ in a book […] [w]e go into a trance-like state that has four aspects. We cease to be aware of our surroundings or our bodies. We tend not to judge the reality of whatever fabulous story or film or play or poem we are ‘lost in’ […]. [W]e feel real emotions toward fictional people and events […]. We know we cannot possibly act to change what we are paying attention to […]. There is an interesting exception: literature in which we do have to act on the work. With the advent of computers came hypertext, in which the reader must continually choose a path through a narrative or poem. Because the reader constantly acts on the work, the experience of being transported becomes impossible. The world cannot evaporate, nor can we feel transported into the world of the story. Instead, we are busy at the computer. I suspect this is why hypertext has never caught on with the reading public. We want that trance-like experience. (Holland, 2009: 40–41)

Note that non-acting does not of course preclude mental engagement with the text, as we shall see notably in the section ‘Why do humans enjoy stories and storytelling?’ subsequently. Indeed, a genre like detective fiction, as well as many literary texts, invites intellectual reflection. Being transported means that we engage with the fictional characters and events as if they were real, inviting deep psychological involvement (e.g. see Gerrig, 1993). Such transportation is only possible if the reader feels able to trust that the psychological conditions provided by the text justify this involvement.

Many commentators have drawn attention to this need to be transported. For example, David Miall (2012), writing from a literary and phenomenological perspective, stresses the need for what he calls ‘absorption’:

[W]hat hypertext sacrifices, through promoting the machinery of reader choice, is the absorption of the literary reader and its invitation to develop the feelings and self of the reader. A hypertext reader, in contrast, must be active in moving about the screen, clicking links, making decisions. The subtle and varying flow of attention typical of literary reading is likely to be thwarted from the outset by the disjunctive structure of the hypertext, with its emphasis on manipulation of screen objects. (Miall, 2012: 203)

If such absorption depends on non-action on the part of the reader, this alone puts hypertext at a disadvantage. A hypertext is not consumed in a form predetermined by its author and functions precisely by expecting the reader to ‘interfere’ in the work. In view of the way this expectation of an active manipulation of the text clashes with the reader’s expectations, it should come as no surprise that hypertext as a genre seems largely unknown to the general reading public.

For most of us, the prospect of an emotionally or intellectually pleasurable experience is the main motivation for literary reading. Hence, the continuing focus in literary studies on interpretation of texts, rather than on the experiences gained from reading them, can be said to fail to acknowledge what is perhaps, for most people outside of academia, literature’s raison d’être (Miall, 2006). Possibly due to their background in literary studies, hypertext theorists doing research on hypertext novels (e.g. Bell, 2010; Douglas, 2000;) ² eschew empirical research of the reading experience, producing instead theorized interpretations focusing on underlying ideology (e.g. gender politics in Shelley Jackson’s Patchwork Girl [Hayles, 2000, 2005]), an alleged effect on the reader, or analyses of single texts in light of some literary (or media) paradigm (Bell et al., 2013). In contrast, empirical studies of literary reading proceed bottom-up, often employing interdisciplinary experimental paradigms in order to determine aspects of literary reading online and offline, such as the degree to which the reader develops feelings for the characters in a story, or the ways in which manipulations of stylistic devices (e.g. point of view) affect readers’ sense of involvement in a text. Theoretical constructs such as immersion (Ryan, 2001), transportation (Gerrig, 1993), engagement (Busselle and Bilandzic, 2009), involvement (Klimmt and Vorderer, 2003), flow (Nell, 1988) or absorption (Kuijpers et al., 2014; Tellegen and Atkinson, 1974) are employed to denote the particular sense of becoming ‘lost in the book’, accompanying enthralling and emotionally pleasurable literary reading experiences.

One influential line of research (Kuiken, 2008; Kuiken and Miall, 2001; Kuiken et al., 2004; Miall and Kuiken, 1994, 1998, 2002) has shown that unique characteristics of a literary text (e.g. foregrounding of stylistic or formal features; literariness) typically elicit four kinds of feeling: (1) evaluative feelings of satisfaction, pleasure, or frustration towards the text, experienced during and/or in retrospect towards the text as a whole; (2) narrative feelings in response to specific aspects of narrative events, characters, plot, and so on; (3) aesthetic feelings in response to striking or unusual stylistic moments; and (4) self-modifying feelings that restructure the reader’s interpretation, thereby prompting the reader to new insights into herself or the world, enabling some transformation, through feeling, of understanding of self.

In one of the earliest empirical studies of hypertext fiction reading, Miall and Dobson (2001) found that the mechanics of hypertext distance the text from the reader, hence discouraging reader absorption and immersion in the literary text. Instead of being transported into the storyworld, the reader is kept busy selecting among links and – often unsuccessfully – attempting to reconcile inconsistencies and lack of coherence between one lexia and the next. However, in this study, Miall and Dobson used a print text, which they manipulated into a hypertext version; hence, their conclusions do not necessarily extend to hypertext novels that are deliberately authored for the hypertext node–link structure. Nevertheless, as they assert, the problems they identified are likely to occur for readers of ‘native hypertext fictions’ since hypertext design by its very nature disrupts the smooth and continuous evolvement of a coherent, textually represented storyworld which is one prerequisite for reader immersion and transportation (Miall and Dobson, 2001). As Miall claims, ‘the short lexias of hypertext […] and the need to choose one from several links may disrupt the reader’s own unfolding dynamic of reading or forestall its development’ (2012: 207).

In another study, Miall (2004) performs a close reading of the award-winning hypertext fiction These waves of girls by Caitlin Fisher, ³ focusing on how it enables the above-mentioned four types of feelings typically prompted by literary reading. Miall suggests that the evaluative feelings are less likely to be of an immersive than of an interactive kind, in which the reader typically finds pleasure in puzzling out relationships between successive lexias. Further, he finds that narrative feelings are likely to be short-lived, as the reader is only provided brief glimpses of characters and settings, affording little time for committed immersion in the storyworld. The ongoing difficulty of establishing the connection between one lexia and the next further frustrates narrative feelings. A third observation is that whilst aesthetic feelings are triggered by Fisher’s work, these feelings ‘are in most cases balked by the jump to a subsequent lexia that fails to develop their implications’. Finally, the fourth and perhaps most important type of feelings in literary reading, self-modifying feelings, are by necessity unlikely: ‘The inability of the reader to sustain a particular focus, to experience a modification of feelings over a series of lexia, suggests that any transformation in understanding beyond the superficial is unlikely to occur’ (Miall, 2004: n.p.).

In addition to the challenges presented by the hypertextual mechanics and the way the narrative is propelled, other studies have focused on how the ergonomic affordances of the hardware (e.g. computer mouse and keyboard) tend to disrupt a reader’s immersion in a storyworld (e.g. Mangen, 2008; Ryan, 2001). These haptic aspects of hypertext reading seem to be largely incompatible with the kind of immersion typically aimed for in literary reading, whereas it is perfectly suited for a different kind of stimulus-driven absorption, namely, computer gameplay (Chaouli, 2005; Mangen, 2006).

Whether or not hypertext novels are de facto incompatible with narrativity is a matter for debate. The direction and outcome of such a debate, however, will largely depend on how one defines ‘narrativity’. Marie-Laure Ryan (2009) outlines the basic condition of narrativity as ‘a sequence of events involving thinking individuals, linked by causal relations, motivated by a conflict, and aiming at its resolution’ (p. 43). Faced with the immense range of possible permutations offered to the reader by the hypertext structure, an author of hypertext fiction has to be content with merely offering building blocks for a narrative (the lexias), which need to be exchangeable across a number of possible sequences (thus contexts) generated by the reader. In other words, node texts (lexias) need to be open ended, and they have to accommodate multiple possibilities for linking. Any resulting narrativity will be constructed largely by the reader. Also, fundamentally, hypertext cannot provide closure, even of the most elementary nature. To what extent these two phenomena fundamentally interfere with narrativity can perhaps be discussed. Ryan, for example, claims that hypertext has a problem creating narrative meaning and immersion in a fictional world due simply to the fact that ‘narrative is a linear, causal sequence of events whose significance depends on their position on a temporal axis’ (Ryan, 2009: 44). At any rate, any narrativity resulting from a hypertext reading is obviously constructed largely by the reader:

If readers are to construct a causal sequence of events out of fragments presented in a variable order, they will have to do so by mentally rearranging the fragments into other configurations than the order in which they were initially presented on the screen. (Ryan, 2009: 44–45)

Fortunately for hypertext authors, the brain’s need for narrative is great enough for the reader to be capable of making up for the lack of narrative guidance, but as the research cited above indicates, the experience may not be as enjoyable as when confronted with a more fully narrated story. The reader’s preferred (mental) activity is more likely to be that of pondering the characters and motives driving the narrative plot than trying to construct a coherent narrative structure in the face of technological barriers. A plot based on chance rather than psychological necessity does not make for a satisfactory experience. There are good reasons to suspect sound evolutionary reasons for this preference. We will explore these in the next section.

Why do humans enjoy stories and storytelling?

As we have seen, a chief characteristic of hypertext is that it leaves the constitution of a reading text (turning an assortment of chunks or nodes into a narrative) to the reader instead of it being the writer’s concern. We have suggested that a major psychological reason why hypertext novels never became a success is that when reading fictional stories, the reader does not aspire to an active role in constituting the text. Rather, the reader seeks to become ‘lost’ in the narrative in a way that presupposes submitting to it passively. We here suggest that the reader’s aversion to taking an active role, such as hypertext demands, is congruent with the way storytelling functions as a form of evolutionarily adaptive behaviour.

It must be noted at the outset that not everyone agrees that storytelling is evolutionarily adaptive at all. Norman N Holland, for example, dismisses all attempts at proving literature to be evolutionarily advantageous (Holland, 2009). Regarding literature from a purely psychological point of view, he believes that providing sheer enjoyment is the clue to its universality and universal popularity. In this sense, he is clearly of the opinion of Stephen J Gould, who has suggested that art (e.g. but also language) may have an evolutionary origin but as a by-product rather than as an intrinsically useful adaptation. In a famous comparison between the human brain and architecture, he likened the way the brain produces art with the way architects produce spandrels: as accidental flourishes arising from a design aimed at loftier purposes, such as survival in the case of organisms and constructional soundness in the case of a building (Gould and Lewontin, 1979). ⁴

Over the last decade and a half, however, the hypothesis that storytelling does function as an evolutionary adaptation has garnered increasing support. Where Holland explicitly pronounces the view that ‘literature cannot train our brain for life’ (2009: 342), the evolutionary adaptation school of thought concentrates on the usefulness of the way storytelling stimulates and enhances the faculty of ‘counterfactual thinking’ (see, e.g. Carroll, 2006; Dutton, 2009; Gottschall, 2012; Gottschall and Wilson 2005; Pinker, 2007; Swirski, 2006; Tooby and Cosmides, 2001; Wilson, 1998). Fictional stories ‘provide a […] complex and useful set of templates and examples to guide and inspire human action’ (Dutton, 2009: 112):

This faculty for imaginative practical reasoning obviously had immense survival value in the ancestral environment, enabling hunter-gatherer bands who were especially adept at it to exploit opportunities, cope with threats, and outplan and outcompete less articulate and imaginative groups and individuals. Fictional storytelling, which likely came later, does not function separately from this faculty but is an enhancement and extension of counterfactual thinking into more possible worlds with more possibilities than life experience could ever offer up to an individual. To the ability to think counterfactually, case-based reasoning adds a capacity to interpret and so gain knowledge by drawing analogies and identifying dissimilarities in richly complex situations that are confronted in reality and contemplated in imagination. (Dutton, 2009: 113–114)

The hypothesis that fictional storytelling serves ‘as an imaginative exploration of the larger possibilities of human intellectual and emotional life’ (Dutton, 2009: 120) or simply ‘mental scenarios’ (Wilson, 1998: 225) is now widely held, even if it has not been proved. As Carroll explains, the ‘vital adaptive function’ served by the arts, ‘including the oral antecedents of literature’ is that of ‘organizing human motives and thus ultimately regulating behavior’ (2006: 41). Acknowledging his debt to EO Wilson (1998), Carroll explains:

In his chapter on the arts in Consilience: The Unity of Knowledge, Wilson supposes that intelligence subserves adaptive flexibility, and he argues that in detaching human behavior from stereotyped instinctive responses, intelligence presented a new adaptive problem – the problem of confusion, uncertainty, and motivational disorientation. The human capacity for the arts, he suggests, evolved precisely as the solution for that problem. ‘There was not enough time for human heredity to cope with the vastness of new contingent possibilities revealed by high intelligence […]. The arts filled the gap.’ (2006: 42)

So the arts in general, but especially storytelling (and its literary expressions), Dutton (2009: 120) claims, serve a particular adaptive feature for humans. Due to our large brains and the complex situations we faced when dealing with other human beings, humans have ‘risen well above the more simple, routinized responses to the environment characteristic of other animals. The fact that human beings in the Pleistocene outgrew automatic animal instincts created problems of its own: confusion and uncertainty in choices available for action’. (Dutton, 2009: 120)

Interestingly, the incapacity/undesirability to act that we encountered in Holland’s psychological explanation of the function of literature is also an essential part of the explanation of the way storytelling might function as evolutionarily adaptive behavior. This fits with the age-old notion of the author who subtly manipulates the emotions of his readers.

[T]he author remains the prime mover, the one who is trying to control the show—the interpretation of characters, their actions, and the events that befall them. Authors attempt this by persuading, manipulating, wheedling, planting hints, adopting a tone, and so forth: whatever will appeal to the reader and create a convincing interpretation. (Dutton, 2009: 125)

This hold over the reader is to a large extent the result of a compelling point of view:

Since individual identity is a crucial feature of the adaptive ecology of human beings, it is crucial also to the construction of meaning in literary texts. Writers are people, and people construct imaginative scenarios in order to satisfy their own psychological needs. The most general such need is the need to articulate and affirm the writer’s own characteristic stances or ways of coping with the world – his or her own beliefs, values, and attitudes. The total set of these beliefs, values, and attitudes constitutes a ‘point of view’, a certain perspective on the world. In this broad sense, there is a distinct point of view implicit in all literary art. Characters in a literary representation, like people in real life, need to affirm their own distinct points of view, but the author mediates among all represented points of view and encompasses them within a single, comprehensive interpretation. The ultimate shaping force behind any imaginative construct is thus the individual identity of the writer. It is for this reason, as Henry James declares, that ‘the deepest quality of a work of art will always be the quality of the mind of the producer.’ (Carroll, 2006: 45)

The cornerstone in the compelling point of view that Carroll identifies is a coherent theory of the mind: ‘What drives the creative process is our hankering for mind-making and mind-reading’ (Zunshine, 2006: 160). Zunshine suggests that it is one of the primary drivers of the act of writing to create a fictional world in which the characters fully obey the dictates of theory of mind. But though fully aware that this creative process deals with fictional characters with fictional minds, the reader, too, expects their motives to obey a sound theory of mind. Not only is it therefore one of the chief sources of enjoyment for the reader to try to understand the psychology of fictional characters but it is precisely in that challenge that the consumption of fictional narratives can be regarded as adaptive.

It appears that the hierarchical relationship between the author and a receptive, passive reader, despised by hypertext theoreticians, is really exactly what the reader of narrative fiction wants and expects. The reader, like the storytelling audience, actually wants to relinquish control and be at the receiving end of the author’s manipulations and wants to be confronted with someone else’s consistently and cogently presented point of view so as to be able to test his or her own theory of mind. If the reader were to act (and a fortiori, if the reader were capable of acting), he or she would be implicated in the narrative events. This would detract from the reader’s sense of witnessing someone else’s experiences, rather than his or her own, and is thus undesirable.

In the storytelling-as-adaptation perspective, as in Holland’s psychological perspective, hypertext thus plays havoc with the expected passivity of the reader, which is needed in order for the ‘transportation’ by an author in charge to take place. In addition, by contrast to the manipulative precision of the author-led narrative, aimed to sweep the reader along by unseen manipulations, reader-joined nodes will necessarily appear to lack cogency and be arbitrary.