Why do we want to read about murder? Most of us do not want to kill people, and most of us would feel a little squeamish if we discovered that one of our friends had done somebody in. Part of the reason must be simple ghoulishness, if it can ever be entirely simple to take pleasure in imagining how people kill each other. In most murder fictions these dark pleasures are overlaid by other, superficially more respectable, kinds of interest. Murder fictions let us test our hypotheses about how and why people act. They may also suggest that the things we do when we read most narratives – forming inferences, seeking intelligibility, constructing hypotheses that fit the plot – might be serious kinds of human activity: they might finally uncover why someone made someone else become a corpse.
The humblest form of murder fiction could be called the howdunnit. This sort of tale is mainly concerned with the causal mechanics of murder, and is more or less straightforwardly an exercise in ingenuity. A man is found stabbed in a Turkish bath. There is no murder weapon, but a tea leaf is found in the wound. Eventually our patient detective works out that the victim can only have been stabbed by an icicle which was smuggled into the baths in a thermos-flask. Howdunnits are the crossword puzzles of fiction. They are elaborately rule-bound, and both author and reader know just how far they can wander towards the arbitrary and the improbable: a discovery that a body was wrapped in clingfilm and moved to a new crime scene in order to establish a body of false forensic evidence is within those rules; a last minute revelation that the murderer used an invisible powder which causes instant death by telekinesis is some way outside them.
Howdunnits are closely related to the much more common and potentially much more sophisticated whodunnit. A good whodunnit works by making its readers think about several possible ways in which things might have happened. The grammatical mood of a whodunnit combines the indicative (this person is dead) with the subjunctive (things might have happened this way), and most of the best of them play elaborate games to ensure that the contingent stories we invent to explain a death take a long time to converge with the bare reality of the indicative mood. A vulgar whodunnit will assemble all the suspects in the billiard room twenty pages before the end of the novel; the detective will present a chain of inferential reasoning; the murderer will turn purple, gasp a confession, swallow a vial of poison, and leave us with a simple indicative truth: (s)hedunnit. A non-vulgar whodunnit might leave a whiff of the subjunctive in the air along with the scent of arsenic: can we be sure (s)hedunnit? Whodunnits might also raise questions about the mechanisms by which we make such inferences, and about the notions of probability we use to enable ourselves to trust those mechanisms; but they deal for the most part in something like truth.
The masterworks of murder, though, are whydunnits. Whydunnits are principally concerned with investigating the motives for and genesis of crime, though they may also worry about the who and the how. Murder necessarily presses at the edges of normal methods of understanding why people do things. For this reason the greatest works of fiction about killing tend to take the question ‘why did he do it?’ and propose multiple and only partially satisfying answers. In Crime and Punishment Raskolnikov kills a moneylender. It’s suggested at various points that he does this because he is poor, because he wanted to be a Napoleonic superhero, and because he may be mad; but none of these reasons seems finally to provide an adequate explanation, either morally or causally, for the action. Raskolnikov finally describes his reasons as though he is improvising: ‘You know, it was like this! This was it: I wanted to make myself a Napoleon, and that is why I killed her . . . Now do you understand?’ To read this particular whydunnit is to watch a variety of motives assemble themselves around a crime at once darker and more absurd than its agent ever recognises. The whydunnit at its best allows large imponderable descriptions of motives (‘he did it because he was free’; ‘he did it because he was evil’) to float around a crime, but it will not let them, or indeed any other single lower-level account of motive, settle over a murder and encompass it completely.
There are two main tendencies within the whydunnit, both of which when taken to extremes lead to bad works of art. The first tendency is towards philosophical triumphalism, where a fiction seeks to demonstrate a particular thesis about why in general human beings act. Because Freud (of course) gave a perfectly satisfactory account of the human mind, Norman Bates must love his mother. Philosophical triumphalism produces the whydunnit’s equivalent of the vulgar whodunnit: it provides a final moment of certain understanding, according to which a particular model of human motive explains the genesis of killing. There may be a bit of readerly resistance (we might find ourselves asking if the crime in Psycho is just a grotesque inflation of the human conduct you would expect to follow from pop-Freudianism, or if finally it’s more about horror in a shower than about the mind of Norman); but that resistance tends to come from outside rather than from within the work itself, as a judgment about plausibility. This kind of whydunnit rapidly looks like a stale piece of the past.
At the opposite extreme the whydunnit leads to something that could be called satirical defeatism. Whydunnits like this amplify the gap between the motives adduced to explain an action and the action itself, and they do so in order to draw attention to the dangers of a dominant view of what makes people act. This sort of whydunnit might also tend towards gothic sensationalism: that is, it might imagine a crime so curiously violent that nothing could adequately explain it. So, in American Psycho, Patrick Bateman kills a girlfriend, apparently because she thought that his suit was designed by Henry Stuart rather than Giorgio Armani. This reduces murder to the merely satirical. A motive and an action are set up against each other in a way that says, without very much intelligence or subtlety: ‘These are the motives which this society says matter and they don’t make any sense; I mean, imagine a guy murdering someone over a designer label.’ American Psycho is murder fiction as cultural warning, and if it weren’t so crude it could be called classically satirical: if this motive or appetite in your society continues to develop, it suggests, then this – lots of killing – is what will happen. Like most satire, it smuggles in a sneaking love of the tendencies it abuses.
There are a few works about murder that display satirical defeatism and philosophical triumphalism simultaneously. L’Etranger is one example: by making Meursault kill the Arab for no reason, or ‘because of the sun’, as he claims, Camus insists in a satirically defeatist way that there cannot be any adequate explanation of this killing. He also seeks to establish a model of human volition – there never can be any adequate explanation of any human action – like the most dogmatic philosophical triumphalist. A complete resistance to all explanations of the reasons people act in a particular way cannot in the end be much more than a thought-experiment, and the fictional versions of such experiments risk turning the complexity of a whydunnit into the endlessly tedious performance of actes gratuits, each equally indeterminate, each equally free, each designed chiefly to unsettle bourgeois assumptions about why people generally do things.
The whydunnits that seem more than period pieces are those which layer one version of motivation over another so as to leave human agency appearing both describable and resistant to simple causal analysis. Tragedies are in this respect the ultimate whydunnits, and the strongest tragedies of murder tend to give many possible explanations for the actions they describe. In Euripides’ Madness of Hercules the gods Madness and Iris drive the hero insane, and he butchers his children. When he returns to sanity he asks: ‘Who has done this thing?’ His father replies: ‘You and your bow and the gods were the cause’ (aitios, ‘the person or thing legally responsible’). Greek tragedy generally offers multi-layered explanations of killing, which tend to entangle human and supra-human agency. This could indicate that the ancient Greeks found killing hard to understand without reference to the supernatural, but it may have more to do with what may be a transhistorical fact about murder: that it cries out for a multiplicity of explanations, the biggest and the best (divine agency) as well as the most irreducible and inescapable – youdunnit. Invoking polycausality in writing about murder may be a cultural self-defence mechanism, which we may have inherited via a number of crooked byways from ancient Greece: by imbuing murder with large numbers of overlapping motives, and by embedding it in fictions which include the occult and the inexplicable (madness, jealous sexual passion, satanic possession, drugs, divine intervention), we tell ourselves that yes, indeed, we couldn’t do it. Murder fictions warn us that killing is the point at which our accounts of human behaviour begin to break down, and at the same time reassure us that we do have a lot of ways of explaining this kind of thing, too many perhaps for any one of them to work on its own: you and your bow and the gods dunnit.
A Cultural History of Causality, despite its ambitious title, has no truck with ancient Greece, or indeed with much of what philosophers or lawyers have said about agency or causality before the 19th century. The book began, Stephen Kern says, as an attempt to understand how the discoveries of late 20th-century subatomic physics influenced modern understandings of human agency. Subatomic particles do not follow the laws of causality laid down by classical physics, in which cause c necessarily precedes effect e. The argument of Kern’s book is, roughly, that as scientific conceptions of causality have become more complex and multifaceted, so our understanding of the world and of the reasons murderers kill people have become less certain and more multiple. This Kern terms the ‘specificity-uncertainty dialectic’: the principle that the more we know and the more specific our knowledge becomes, the less we understand and the more we see remains to be known.
Kern says he does not like ‘crudely self-congratulatory Whiggish history’, but his history of murder fiction since 1830 does seem decidedly Whiggish, albeit Whiggish in a form which ends in the triumph of uncertainty rather than of reason. He argues that Victorian novelists had relatively simple linear and determinist conceptions of what made people kill: degeneracy, evil ancestry, simple jealousy, telegony (the belief that children by a later sexual partner could display characteristics of an earlier one). The moderns (by which he means anyone writing after about 1890) tend to fracture and multiply these causal structures, and to favour non-linear models of causality which depend on the unpredictable operations of complex systems. There is no doubt that Kern regards the moderns as wiser than their ancestors because they have a more precise and various understanding of the world: so Angel Clare’s despair on learning of his wife’s sexual history in Tess of the D’Urbervilles ‘is a culmination of the anxiety about impure blood and hereditary taints that plagued the Victorians in their ignorance of genetics’, while Modernist novelists ‘had a more specific and precise understanding of jealousy than Victorian novelists had’. What he doesn’t fully reckon on is the likelihood that every age thinks its understanding of why people do things is more complex than that of its predecessors: Zola thought he knew better than Fielding and Robbe-Grillet thought he knew better than Zola. We moult old explanations for human conduct and sprout new ones, overlayering the new and the old. Gradually societies come to see as reductive and thin explanations that their ancestors saw as rich and perplexing. Disowning the dominant views of human motive from a previous age is the way we persuade ourselves we’ve moved on, but fiction often reminds us that moving on is not easy, or that as we move on we move in more than one direction.
Kern loves a boffin, because boffins are very keen to tell us we have indeed moved on, and the book gives some strong, clear and interesting expositions of scientific and cultural theories. But this virtue does not make up for its two central defects, which both, oddly enough, arise from its conceptions of causality. The first is its use of ‘cause’ to describe any attempt to explain human behaviour by reference to anterior circumstances. According to this view, sociology offers a model of causality; telegony and theories of degeneration offer another; subatomic physics yet another; not to mention greed, theories about language, sexual jealousy and discoveries about hormones. I would balk at calling most of these ‘causes’ of crime: they are rather reasons which might retrospectively be invoked to explain murder. Dostoevsky doesn’t suggest that poverty causes Raskolnikov to kill the money lender, and he isn’t willing to give unironised assent to Raskolnikov’s claim that ‘I wanted to make myself a Napoleon.’ When we are asked to describe why a person did a particular deed, we are not being asked a question about the necessary and sufficient prior conditions for that deed. Indeed, causality narrowly considered is the preserve of the forensic howdunnit: the cause of a murder-victim’s death is, from this perspective, a haemorrhage which resulted from a blow on the head which resulted from a blunt instrument. The whydunnit begins when we start to think that the person responsible for bringing the blunt instrument down on the victim’s head might have a motive (lust, greed) or a story which could serve to explain the action. The murderer may have wanted to be Napoleon, or may have a deranged love of his mother, or may have decided that morning to kill someone. Rolling a mess of explanations into a ball, connecting them with the bare causal mechanics of murder, and persuading you that they could make someone perform an act at the boundaries of the imaginable is one part of the whydunnit; its other aspect is an aesthetic challenge: can anything explain why someone should, as the hero of Thomas Harris’s Red Dragon does, kill and rape his victims, insert shards of broken glass into their eyes and elsewhere, and then bite them with his grandmother’s false teeth? Wonder at the possible collapse of ways of explaining has always been a major part of the whydunnit’s appeal.
The second big thing wrong with the book is its extremely deterministic view of the relationship between changes in systems of thought and changes in ways of writing about killing. For Kern, books about killing more or less do what scientists tell them to do. Physics gives us the uncertainty principle, so we get Don DeLillo and Thomas Pynchon. Lamarck and Darwin give us Hawthorne and Hardy. Sometimes, though, novelists let us down, so Kern has to rap their knuckles for representing things which scientific studies can’t corroborate: ‘The demands on murder novelists to provide tidy concluding explanations has led them to imagine horrendous childhoods to explain their characters’ actions, but researchers have found no significant correlation,’ he solemnly records. Somehow I don’t think he’s quite grasped the point of mythologies. He also spends a lot of time telling us about the contribution of neuropeptides to the experience of fear, and how researchers have found a relationship between some kinds of hormonal imbalance and some forms of violent behaviour. He is obviously disappointed that novelists have failed to use these discoveries to motivate their heroes. But could a modern Raskolnikov say: ‘This was why I did it: the peptides which control the level of greed in my body were at dangerously low levels, and owing to abnormalities in my hypothalamus my serotonin levels were low. Now do you understand?’ The reason he could not is that a murder story can only deploy a model of causality which is assimilable to ordinary language descriptions of human agency, or which could seep into the language of the law. Our legal discourse of murder could admit a claim that the killer had abnormal hormone levels as part of a plea of mitigation, or as part of an attempt to reduce a charge of murder to one of manslaughter. But an outright causal model of brain chemistry takes us too close to philosophical triumphalism to be something a novel could be quite comfortable with on its own, or without some degree of irony surrounding it. Murder stories can only feed on models of causality which are sufficiently inadequate to make the agent’s deeds seem something short of inevitable.
It is probably true that whydunnits in the 20th century were more prone than those of earlier periods to describe human motives in consciously multiple or irresolvable ways. But Kern does not seem to allow that fiction in earlier periods could be sceptical, even programmatically sceptical, about dominant theories of why people do things, or that much of the point of narrative fiction which foregrounds questions of agency is to explore the various and contradictory ways you can explain a deed. There is a greater rule in fiction and cultural history than the specificity-uncertainty dialectic: no story which invokes only one model of causality to explain human actions could ever be worth reading except as a freakish experiment. That was true in 400 BC, true in 1850, and is likely to remain true. The reason, I suppose, is that we turn to fiction, and especially to murder fiction, to help us see what we don’t understand.