MORE THAN 130 TELEVISION stations in the United States label their local news broadcast as “Eyewitness News.” Or so says Wikipedia, which tends to be good evidence for this sort of thing, even if not for everything. And because the label designates a style of local news reporting featuring camera crews rushing to the scene of newsworthy events and reporters breathlessly interviewing participants in ongoing happenings, the image these stations wish to project is that of the reliable firsthand witness.
But just how reliable are eyewitnesses? More precisely, often we want to know how reliable are the representations, as evidence, of those who have claimed to have seen something firsthand. Or, less commonly, to have heard something firsthand. Or, even less commonly, to have tasted or smelled or touched something firsthand. The question, the answer to which might not be the same for hearing, smelling, tasting, and touching as it is for seeing, is not just about lying, which we have explored in Chapters 6 and 7. Rather, it is about the even more common phenomenon of honest mistake by those whose firsthand accounts we are asked to believe just because of their firsthandedness—just because they were there.1
The reliability of eyewitnesses—or, more accurately, the unreliability of eyewitnesses—has become prominent in recent years.2 Eyewitness accounts have always been a pervasive source of information and knowledge, but much of the recent prominence is due to the subsequent exoneration of people who have been convicted of crimes they did not commit. It turns out that the largest number of these exonerations have involved trials in which an erroneous conviction was based on what was subsequently discovered to have been a mistaken eyewitness identification of the unfortunate, actually and factually innocent defendant—the defendant who simply didn’t do it.3 Such mistaken eyewitness identifications, however, come in three varieties, which we need to disaggregate. The first is the one most people imagine when they think of eyewitness mistakes—a failure of perception. For example, believing you saw a bird when what you really saw was a bat. Or believing you witnessed a shoplifter when the person leaving the store with the goods had actually paid for them. Or believing you saw Chris when you really saw someone else of Chris’s approximate height and weight dressed the way Chris dresses. Second is a failure of memory—perceiving something accurately at the time of perception but remembering that perception inaccurately some days, weeks, months, or years later.4 For example, knowing at the time of witnessing an automobile accident that the blue car that ran the stop sign was a Toyota but (mis)remembering much later that it was a red Honda. Or knowing when you were growing up the telephone number of your childhood home but mistakenly thinking now that it was a different one. And third is a failure of reporting—inaccurately reporting what you accurately saw and now accurately remember. An example would be describing what you accurately saw and accurately remembered as sleet in a way that your listeners believed that you were describing snow.
All these failures, and more, have produced a healthy skepticism in recent years about the value as evidence of eyewitness accounts. And much of this skepticism has had admirably important effects on the legal system. The procedures for identifying a perpetrator in a lineup, for example, have been changed substantially in recent years to lessen the possibility of a mistaken identification caused by manipulative law enforcement tactics designed to steer the identification in the direction of the individual that the police had already decided had committed the crime.5 In addition, skepticism about criminal convictions based solely on the testimony of one eyewitness has produced the further investigation that frequently reveals that a conviction and subsequent incarceration were based on a demonstrable factual error. Mistaken eyewitness identifications, sometimes revealed by credible recantations by the single witness, have, it turns out, produced the single largest number of post-conviction exonerations. Indeed, the risks of mistaken identification, even by those who claim to have observed the defendant actually committing the crime, are sufficiently substantial that some states, starting with New Jersey, now explicitly instruct jurors about the possibility of error in an eyewitness identification, a possibility that the typical lay juror is inclined to ignore or minimize.6
Even outside of the legal system, honest mistakes by those who claim to have observed something “with their own eyes” are widespread.7 And so too with people who insist they have remembered something, even though mistakes of memory are as predictable as mistakes of perception.8 There is thus good reason to doubt that eyewitness accounts of past events are evidence that we should treat as invariably reliable just because it is provided by people who purport to have seen or even experienced firsthand what they are now reporting to us.
But recall the lesson of horses and zebras from Chapter 2. It is true that neither memory nor first-person perception are as accurate as they have traditionally been understood to be. And it is important that we know this, especially because of the consequences in the criminal justice system and occasionally elsewhere of the overvaluation of first-person or eyewitness accounts. Still, most—indeed, a lot more than most—eyewitness accounts are accurate, and as such can be evidence of what the eyewitnesses claimed to have seen. And most memories are accurate as well, providing, again, evidence of what those describing what they have seen, heard, or experienced have in fact seen, heard, or experienced. We all have a list of things that we have remembered inaccurately, but even a long list of inaccurate recollections pales in the face of what for almost everyone is a far longer list of accurate recollections. I may have forgotten what I had for breakfast yesterday or where I left my keys today, but I remember, correctly, a very large number of other things about yesterday and today. That both perception and memory are often flawed is not inconsistent with perception and memory being even more often accurate and thus of evidentiary value. Even though eyewitness accounts have traditionally been overvalued as evidence, it is possible that recent revelations about this overvaluation, and its consequences, have led to substantial undervaluation.
One cause of what may well be the contemporary undervaluation of eyewitness accounts is, as just noted, the common tendency to overcompensate for recently discovered flaws of one sort or another. Disappointed believers often become radical disbelievers, and some of the contemporary skepticism about eyewitness accounts may be a function of just this kind of overreaction when people who used to believe that eyewitness accounts were always or almost always reliable become aware of how such accounts may be mistaken. The publicity about exonerations in the criminal justice system has shone valuable light on the consequences for individuals of the traditional excess reliance of eyewitness accounts generally and eyewitness identifications in particular. But it is important to remember that although a mistaken identification in a criminal prosecution can result in the incarceration or even execution of someone who is innocent, not all mistakes in first-person accounts have the same dire consequences. Recalling Blackstone and the way in which the errors of mistaken conviction are properly treated as much more serious than the errors of mistaken acquittal, it is entirely appropriate to treat the mistakes of an inaccurate eyewitness account in the criminal justice system as especially grave, and accordingly to be willing to suppress or discount some number of accurate eyewitness accounts in order to avoid a smaller number but more serious inaccurate ones. We do not know how many guilty people would have been acquitted had juries been (properly) instructed to be skeptical about eyewitness identifications, but it would be surprising if that number were zero.9 An attitude or disposition of skepticism about eyewitness accounts, being of necessity general rather than event-specific, will be skeptical of accurate eyewitness accounts as well as inaccurate ones, with the consequence being that some number of accurate eyewitness accounts will be mistakenly discarded as inaccurate. Moreover, and as we saw in Chapter 3, coming to believe that an eyewitness account was not beyond a reasonable doubt accurate is consistent with it being probably accurate—even though, if that is the only or primary evidence, it is appropriately insufficient for a criminal conviction. And that is why at least some—we do not know how many, but it would be remarkable if it were none—of those who, on the basis of recent revelations and research about the failings of eyewitness identification, have been properly found not to have been guilty beyond a reasonable doubt may well have been probably, but only probably, guilty of the crime of which they were charged.
Once we leave the criminal justice system, however, Blackstonian skewing is by no means the only or the necessarily correct approach to any question about how we should treat eyewitness accounts as evidence. Treating the use of mistaken eyewitness accounts as more serious than the non-use of accurate accounts is what the law ought to do when it is prosecuting people for alleged crimes and threatening them with incarceration, or worse, but that conclusion comes at a cost of overall accuracy. When we exit the criminal courtroom, the same sacrifice in overall accuracy may no longer be appropriate. Consider the question whether I should return to the restaurant where a year ago I enjoyed the best chocolate soufflé I have ever had. My recollection of that exquisite soufflé leads me to want to return. But perhaps the soufflé was not as good as I now remember it. Or perhaps it was a different restaurant. Or perhaps it was a cheese soufflé and not chocolate. Or maybe it was chocolate mousse and not a chocolate soufflé. All these things are possible, and the awareness of false memories and maybe even false perceptions alerts me to such possibilities. But there are worse things in the world than ordering a disappointing chocolate soufflé or going to a restaurant in the expectation of having one and discovering that it is not on the menu and never has been. Being imprisoned for a crime you did not commit, for example. And although the chocolate soufflé example may be extreme, it illustrates that mistakes of perception and of recollection might be more serious in some contexts than in others. As a result, it is wise to remember that although mistakes of this kind are always possible, they are rarely probable. And the larger lesson, which pervades all the chapters here on testimony, is that although it is correct to be somewhat skeptical about testimony as evidence of what the testimony asserts, or at least to be aware of the possibility of false testimony, it is not necessarily correct to treat an item of testimony as having a lower probability of truth than it actually has.
Much of what was in the foregoing pages became highly relevant during the Senate trial of the impeachment of Donald Trump in February of 2021. Perhaps ironically, and perhaps hypocritically, much of the recent concern about the accuracy of eyewitnesses and first-person accounts appeared to go out the window as the impeachment proceedings in the House of Representatives and then the trial in the Senate unfolded. Some members of Congress saw little need for witnesses or other evidence, and that was because they were there. They could make their judgments, these members of Congress alleged, on the basis of their own experiences and their own observations, but with seemingly little recognition of the possibility that the very traumatic nature of those experiences would have rendered their observations more suspect rather than more reliable. And so too for many of the rest of us, who saw the events on television as they were happening and then multiple times thereafter. If eyewitnesses are to be distrusted, are all of us to be distrusted in terms of what we saw on television on January 6, 2021?
The answer to that question is not straightforward. For one thing, an impeachment trial is not a criminal trial, and no one is going to be imprisoned or executed solely because of the Senate’s verdict. Accordingly, it might be appropriate to rely more on our commonsense confidence in observation even though it can (fairly rarely) go astray. In a criminal trial, with imprisonment or even execution looming, it is fully justified to guard against the small but not nonexistent possibility of mistaken first-person observation, as the New Jersey courts have led others in doing. But in a proceeding other than a criminal trial, perhaps relying on the likely correctness of an eyewitness observation makes more sense than guarding against its unlikely incorrectness. Nevertheless, it is a point of some amusement or irony that many of the most skeptical voices about the dangers of relying too much on eyewitnesses appeared strangely silent with respect to the most eyewitness-dependent event in recent memory.
There is a tedious consistency in the opening lines of the letters to the editor that most newspapers print on or near their editorial pages. Although the letters deal with all varieties of subjects and express all varieties of points of view, a remarkably high percentage of them begin in the same way: “As a …”
“As a retired firefighter, …” “As an earthquake survivor, …” “As a psychiatrist who has treated many cases of claustrophobia, …” “As a victim of a hit-and-run accident, …” “As a college student, …” “As a recent immigrant, …” “As the descendant of grandparents who died in the 1918 influenza epidemic, …” And so on and on and on.
What is interesting about the “as a …” claim is that the reader is asked to treat what the letter writer says as more reliable—as better evidence—because of the letter writer’s first-person experience. Sometimes the “as a …” claimant is, plausibly and often usefully, engaging in an act of self-credentialing, as with, “As a former winner of a Nobel Prize in medicine, I believe that the most effective way of controlling the Covid-19 pandemic is …” Or “As the author of seven books about climate change, I have become convinced that the principal cause of the melting of the polar ice cap is …” And sometimes the self-credentialing will not be about expertise, but about an opportunity to perceive that is likely better than the reader’s. “As someone standing outside the Capitol building on January 6, 2021, …” “As someone who was working in a bank on Wall Street on September 11, 2001, …” “As someone who once ate a home-cooked dinner at Julia Child’s house, …”
Such self-credentialing is common, understandable, and often useful for the reader. But if we set aside the self-credentialing that describes the writer’s source of genuine expertise or a genuine advantage in perception, we see a different variety of the “as a …” claim as especially common. Here the letter writer is not claiming the authority of professional or expert knowledge, and maybe not even the ability to observe that others lacked, but instead the authority of the survivor, the participant, the victim, or some other form of first-person experience. And then the question, plainly related to the issues about eyewitness identification that we dealt with in the previous sections, is the extent to which someone’s participation in some event gives them some sort of special authority or special insight into that event solely by virtue of their participation, such that we should treat their account as evidence having significant weight.
Now things become more complicated. Sometimes the authority of the person who was there is based on an opportunity for perception that the receiver of the testimony probably did have—for example, the person who was at the Capitol on January 6, 2021, and who may have been able to see something that the rest of us could not. And sometimes being a participant provides the motivation to learn more. A person who is afflicted with a disease might be motivated to read more about it, and to follow the research more closely, than others whose interest in the disease is less personal. Members of Mothers Against Drunk Driving (MADD) whose children were killed by drunk drivers have profound reason to learn more about drunk driving, its causes, and its possible cures than do those of us not so personally affected.
All of these qualifications duly noted, it remains the case that it is common to give some kind of special testimonial, and thus evidential, weight to participants solely because they were participants. Yet it is not clear that simply being a participant is necessarily a source of greater knowledge or greater insight. For one thing, being there can be distorting as well as enlightening. A driver injured in a minor traffic accident is not only likely to have a particular view about who or what caused the accident but may also see the accident or the injuries as more serious than they would appear to a less involved observer. And the same applies to being a victim more generally. People with afflictions sometimes caused by environmental agents—toxic waste dumps, lead pipes, polluted water supplies, and such—often have opinions about what has caused their affliction, and often they attribute the affliction to some newsworthy toxic agent. Frequently those opinions are correct, but they are not correct by virtue of the affliction being theirs and not someone else’s. People may know the cause of their broken arms, but the fact that hair loss is sometimes caused by secondhand smoke does not give the person whose hair is falling out any special insight into whether secondhand smoke was the cause, despite how frequently victims offer such opinions on the evening news.
Quite often someone claiming the authority of first-person experience talks of having a particular perspective. The former member of Congress does indeed have a perspective on Congress different from those of us without that experience. But does that perspective confer value as evidence on what the person with that perspective has to say? The perspective of the insider is indeed a perspective not shared by outsiders, but outsiders have perspectives too, and the outsider perspective is not shared by the insider. Does the insider perspective, by virtue of its insider-ness, confer more reliability or more reliability as evidence for some conclusion? That depends on what the conclusion is, or what the insider claims to know that non-insiders do not.
One of the most profound observations about the insider perspective was offered by Izaak Walton in his 1653 The Compleat Angler. Here Walton warns the reader that no advice from a fisherman on how to fish would be worth considering if the fish could talk. And although I think I can tell you about university teaching from my perspective, there is little reason to believe that this perspective is more valuable than that of the students, or, even better, the clients my law students will eventually serve.
This idea is familiar to many of us from the Indian parable about the blind men and the elephant, most familiar these days from the 1873 poem of that name by John Godfrey Saxe. As the parable goes, a group of blind men are taken to “see” an elephant, but, being blind, they cannot actually see the elephant. So each touches a different part of the elephant, and generalizes—analogizes—what an elephant is like from the particular part that he happens to touch. One touches the trunk and imagines an elephant as being like a rope. Another touches the leg and believes an elephant is like a tree. Still another touches the elephant’s side and thinks an elephant is like a wall. And so on. And if there is a lesson from the parable, it is that partial perception is just that—partial.
The parable of the blind men and the elephant holds a valuable lesson for thinking about the authority of first-person or direct experience. Participants, perpetrators, victims, bystanders, and others—all others, in fact—have a standpoint, or a perspective, or a lens through which they view an act, an event, a state of affairs, or anything else. And although relying on the accounts of eyewitnesses may be more valuable as evidence than some of the contemporary skepticism might warrant (or so I have been arguing here), relying on the accounts of participants may be less valuable than some of the contemporary celebrations of so-called lived experience might warrant. Insofar as some perspectives have traditionally been ignored, that is a huge problem, and one in desperate need of remedy. Plainly some of the call to recognize those experiences and the feelings they generate is the morally generated and morally justified plea to stop ignoring the perspectives of those whose perspectives have long been ignored.10 But those perspectives are still perspectives, and their value as evidence needs to be understood as such. The accounts of those who say, “It happened to me,” just like the accounts of those who say, “I was there,” have value as evidence, but that value, which is to be distinguished from the moral values of inclusion, may, as with eyewitness testimony, suffer from the flaw of being overvalued as evidence—just as first-person accounts often suffer from being morally undervalued.
Back when all cameras used film (or, even earlier, when they used glass plates), photographs were understood to be especially reliable evidence of what they were photographs of. Not many people know what Mount Everest or an emperor penguin looks like, but they think they do because they have seen pictures. And the same for Franklin Roosevelt, Charlie Chaplin, and Jacqueline Kennedy, all distinctive-looking people who were photographed countless times but were never the subject of memorable paintings or drawings.
For those of us who claim to know what those people, places, and animals look like, our evidence for that knowledge comes entirely from photographs. And like the evidence that comes from eyewitnesses, photographic evidence has long been understood to be particularly reliable evidence, even if not perfect evidence, of the characteristics of the people, places, and events that the photographs are photographs of. The photograph has long been understood as being like the report of an eyewitness, only better.
Susan Sontag, among others, knew better. And so do most serious photographers. In On Photography, Sontag demonstrated just how much a photograph was the product not only of what the photograph was a photograph of, but also of numerous choices made by the photographer.11 Most obvious is the choice of what to leave in and what to take out, not because of any fancy darkroom or Photoshop or retouching techniques but simply by virtue of where one aims and where one does not, or how close or far away the camera is from what it is that the camera sees. Consider, for example, a still-discussed episode from the 1959 Boston mayoral election, in which the underdog, city councillor John Collins, defeated the heavy favorite, John Powers, who was then president of the Massachusetts state senate. Powers had long been reputed or rumored to be associated with gamblers and other organized crime figures, and Collins capitalized on this by using in his campaign brochures and making available to the newspapers a photograph of Powers with a particularly notorious bookie. The Collins campaign, however, had cropped out of the same picture a third individual, the revered Cardinal Richard Cushing. Being associated with a known gambler had bad implications, and being associated with Cardinal Cushing had good ones, so the cropping, which may have helped Collins in this election but which dogged him for years afterwards, made what was originally an innocuous photograph look far more sinister than it actually was. And those who saw the cropped photograph were none the wiser.12
As with eyewitness accounts, the view that it’s a lie to say that “the camera doesn’t lie” can be taken too far. Noticing the frequent errors made by eyewitnesses can blind us to the fact that most eyewitness accounts are mostly right. Similarly, recognizing that a photograph reflects the choices and perspective of the photographer—as well as the results of optical, chemical, and, these days, digital processes—blinds us to the fact that a photograph is usually pretty good evidence of some aspect of what the photograph is a photograph of. The same holds for what the television camera shows—the storming of the Capitol on January 6, 2021, being the most salient recent example. A philosopher of art, Kendall Walton, reminds us that a photograph is more than just a depiction or a description. It is importantly “transparent” to what is being photographed in the way that seeing something through a microscope or telescope is different from the descriptions in prose and poetry and depictions in drawings and oil paintings.13 Of course photographs can be distorted and even fabricated. The government of Nepal, for example, has just taken steps to revoke the climbing permissions of two mountaineers whose photograph of themselves on the top of Mount Everest was exposed as a complete fraud.14 And even non-fabricated photographs reflect the standpoint, the perspective, the theories, and the values of the photographer as well as the mechanical, optical, and, now, digital properties of a human-created device. But there is a reason newspapers give us photographs and not drawings, why television news often seems more informative than radio news, and why Matthew Brady’s Civil War photographs tell us something that the roughly contemporaneous oil paintings of Édouard Manet or the drawings of Francisco Goya a generation earlier do not.15 As evidence, photographs are far from perfect and far from “pure.” But their value as evidence is different from and often better than other means of conveying much the same information.
A photograph is thus best understood as testimony, of a mechanical variety. As with all testimony, it might be distorted. It might be incomplete. Depending on the skills and motivations of the photographer, it might be a complete fabrication. Simon Brann Thorpe’s war photographs use only toy soldiers and fabricated landscapes. For Thorpe, photographing what is itself artificial emphasizes the creative role of the photographer and downplays the role of that which a photograph is a photograph of.16 Just as we have come to understand that not everything that someone reports is true, we have also come to understand, assisted by commentators like Sontag and photographers like Thorpe, that not all photographic testimony is accurate. But here we are back to the question “Compared to what?”
The legal system’s approach to photographs reflects the law’s traditional and at times peculiar preference for oral testimony. So, traditionally, a photograph will be admitted as evidence only if someone—not necessarily the photographer—testifies that the photograph accurately represents what the photograph is supposed to be a photograph of.17 And although photographs authenticated and used in this way are typically referred to as “pictorial testimony,” the photograph is not so much testimony as it is the adjunct to someone else’s testimony—in the manner of a chart or a graph. A photograph used in this way is thus analogous to the witness who holds out her hands in support of her testimony saying that the person she saw was “this tall.”
More recently, and spurred by the proliferation of hidden and not-so-hidden cameras that used to be situated only in banks and are now located almost everywhere, photographs taken by such devices can be used as evidence—as a “silent witness”—as long as there is testimony by the installer, operator, or someone else who can testify that what the camera photographed was what the camera saw and what was actually “there” when the camera recorded the image.18
Plainly most of the considerations that apply to photographic testimony apply to nonphotographic sound recording as well, and to videotape in all of its permutations. Indeed, the law again has had to wrestle with the problem of videotapes that portray one side’s version of the events with such seeming realism that jurors are likely to believe that they are seeing what actually happened, as opposed to one side’s version of what actually happened.19 But through all of these technological developments, from Louis Jacques Daguerre’s earliest cameras to the most contemporary forms of digital imaging, from Thomas Edison’s first phonograph to state-of-the-art electronic recording, the issues are the same, even if they vary in degree.20 All of these techniques can convey what they record and are thus analogous to human beings who convey via testimony what they have seen, heard, or otherwise experienced. But like human beings, photographs and recordings also can lie, palter, fudge, hedge, slant, and embellish. The question, though, is not whether they can. The question is whether they do. And just as a slanted story often supplies more information than no story at all, a photograph can and typically does provides more and better information than a verbal description or a drawing, both of which typically involve a substantially higher ratio of human creativity and intervention than technological reproduction.
Perhaps the best analogy here is to the locks on our cars. Even in my small city of 50,000, I am pretty sure that there are a few hundred people who in a space of just a few minutes can unlock my locked car and drive it away without benefit of a key. But I lock my car anyway, because I recognize that what can happen usually does not, and because I recognize that making things more difficult decreases the likelihood of theft, even if it does not eliminate it entirely. Similarly, although it is possible that the photographs of the assault on the Capitol on January 6, 2021, were complete fabrications, and even though it is probable—actually, certain—that they reflect the perspective, in the broadest sense of that word, of the photographer, they are not only far better than nothing—they also give an impression that is undoubtedly highly accurate, even if not perfectly accurate. And so too with the video recording of police officer Derek Chauvin’s killing of George Floyd, and, even earlier, the video of football player Ray Rice assaulting his wife in an elevator. In cases such as these, it is possible that the perpetrator’s denials would have been believed, but the photographic evidence added such a degree of reliability that flat-out denials became implausible. And thus both incidents are valuable reminders that although photographs and videos can be distorted, misinterpreted, and occasionally even fabricated, they often serve as far better evidence than first- or secondhand testimony, which can also be distorted, misinterpreted, and fabricated, and often to an even greater degree.
The lessons about photography, therefore, track the lessons about eyewitness or other forms of first-person testimony. Skepticism is justified, but complete skepticism is not. Of course we should recognize that the inevitable human choices are as inherent in photography and sound recording as they are in human descriptions. But if the right amount of skepticism produces the conclusion that photographic evidence is far from perfect, we should remember that evidence that is far from perfect—just like almost all other evidence—is far from worthless.