Session 2: Establishing the Features of the Consumer
The UK courts have in recent years been quite explicit that the consumer is a normative construct, a fiction, and a benchmark. What does this mean? Does the modifier “average” or “reasonable” serve any useful purpose? Do courts elsewhere understand the concept of the consumer similarly? If the concept is “normative,” from which materials or domains is this normative construct pieced together? From where ought these norms to be derived? And why force those normative concerns into the vehicle of the consumer rather than vindicate them explicitly on their own terms? To what extent and in what trade mark institutions (legislators, administrative offices, courts etc.) is there a role for empirical understandings of the consumer? If there is a role for empirical understandings, how are these to be developed and used? Through surveys? Courts in several countries appear increasingly sceptical of survey evidence? Does this reflect (or will it cause) a different conception of the consumer? Through other forms of evidence from experts in marketing, economic, behavioural sciences? If there is a role for empirical understandings within a normative framework, what is the relationship between empirical and normative assessments? Must one always trump the other? Does a “trump” only work in one direction (e.g., in favour of trade mark owner or third party user)? Does this vary by institutional setting? Is the consumer construct different for different categories of trademark subject matter—e.g., for verbal versus visual marks?
Introduction: Jessica Litman: US judges and lawyers think what they’re doing is directed at what actual consumers perceive, but the Q is how you figure that out. Corollary: consumers differ by product. Real consumers don’t distinguish TMs from products. Trying to figure out what the consumer is confused about when perceptions don’t track legal categories is challenging. US judges reject surveys a lot, maybe even more and more, not because they think surveys aren’t relevant but b/c they think the surveys don’t reflect empirical reality. They are lawyers and know what they’d do if they were litigating; getting a realistic survey is not good litigation strategy. Some judges are trying to figure out policy for competitive space; starting to see move away from multifactor likely confusion test, especially in the 7th Circuit—this is or isn’t going to confuse the relevant consumer. Sense is that’s also cropping up in false advertising cases—doctrinal categories that require surveys & judges worm their way out of those categories with, e.g., falsity by necessary implication.
Are there ways to generate more reliable empirical evidence of consumers’ impressions? Independent surveyor is interesting idea. If consumers can’t tell difference b/t product and brand, would we be better served by a more careful definition of what sort of confusion ought to be actionable?
Kimberlee Weatherall: Must be frustrating to have judge ignore evidence on which you would rely as a business matter. When you know you have to convince the judges about closeness, and that’s what it takes to win, that inherently leads to uncertainty that makes settlement more difficult; hard to predict how judges will see marks. Intellectually dishonest to reject surveys: expect level of perfection from surveys not expected from anything else. (Preach!) Other experts; witnesses who testify they were confused, as if witness prep didn’t happen.
But TM law doesn’t ask empirically testable questions: a consumer with a fictional memory for the mark as registered, and then we ask about what the consumer will think of the product’s origin when she sees it. But lots of times the consumer isn’t thinking about product origin at all, but about what taste of ceral they want or what the 3-year-old is pulling off the shelf. Fictional memory, imaginary state of mind, imagined shopping experience—and then we complain that we don’t have ecologically valid evidence! We’re not asking testable questions; we assume in the law a kind of decisionmaking that cognitive scientists don’t agree w/ in their model of consumer decisionmaking: conscious and unconscious elements of choice that vary across circumstances.
Courts have rejected general testimony about consumer behavior as applied to specific circumstances; not good. She’s been working with cognitive psychologists on word similarity, memory, etc. Asking psychologists how to test some of our assumptions about marks—e.g, beginning of words mattering most; consumers focus on most distinctive part of mark and ignore generic/descriptive elements. Process is difficult b/c we don’t speak the same language, and they break down the process of consumer decisionmaking differently. We wrongly tell researchers the question to ask and we perhaps should give them scenarios and ask what could be empirically tested: get them to design surveys. But raises practical Q: will a court accept such abstracted testimony as relevant? (RT: by ordinary relevance standards it really should be accepted!)
Primary Discussant: Annette Kur: Consider whether we want to influence how consumers react w/law—miss out on that if we concentrate on making surveys function as well as surveys can function. We can’t be under the illusion that we can get all we’re interested in doing into one litigation. Complexity of problem might justify giving up trying to get empirical evidence in any particular case, except for secondary meaning.
Registration: if it’s difficult to bring in empirical evidence to litigation, even more impossible when we talk about registration. But we still need rules. Relative grounds for refusal: attempt to get as close as possible to actual consumer reaction is illusory and simply should be abandoned.
Robert Jacob: Doubts about so-called experts. People who come along and tell you they’ll give expert testimony on royalty rates etc. are legion, but they’re rubbish. It ought to be true for TMs: if it’s that complicated, you haven’t got a TM case.
David Llewellyn: Weatherall’s point about lost bearings w/reality is worth reinforcing. He’d hate having a decent judge abdicate responsibility to experts, who can be appalling. Generalists should have the nerve to make their own judgments. SMEs: all this survey evidence can be achieved by Pepsi and Coke, but that’s an expensive area; we can all design fantastic ways of getting answers with an open checkbook, but how does the small business w/ a registered TM actually compete? We don’t have contingency fees. Cross-examination and real witnesses are what’s important for a case—real differences in procedure matter, such as presence of contingency fees.
Weatherall: there are empirical claims you can make as a general matter, such as whether the beginnings of words are most important. Also one reason for the huge expense is b/c courts have set the bar so high. A cheap internet survey may give you more evidence than you had.
Llewellyn: If you pay peanuts you get monkeys; what kind of legal system do you want to have?
McGeveran: distinguish between consumer as beneficiary (as all law is supposed to benefit the public) versus consumer as determining scope/lodestar. A judge who doesn’t feel comfortable is being told this is a complex factual dispute with a multifactor test; that kind of judge—our greatest worry—will be tempted to use a survey. Maybe we should be more explicit about being normative.
Grynberg: courts of appeals are much more comfortable with district courts making those kinds of calls than w/agencies. Example: TTAB’s Pretzel Crisps decision: Fed. Cir. says that TTAB shouldn’t have discounted the empirical evidence/surveys, even though the TTAB deemed them unpersuasive and self serving.
Dinwoodie: w/o empirical tools, you either end up w/judge as your empirical proxy or judge as normative decider. Which is the consequence of rejecting surveys?
The extent to which courts appear more persuaded by individual entirely unrepresentative witnesses than scientifically conducted but imperfect surveys: why? Maybe it’s procedural—cross-examination, understanding the thought process of an unrepresentative consumer. (Or maybe it’s just a standard cognitive bias towards vivid examples.)
Jacob: real consumers in the box are perfectly normal people. Seeing them makes it easy to believe them if they seem competent.
Dinwoodie: getting a grounding in empirics could be presented to judges and policymakers in other fora; wouldn’t need to be presented in particular cases but as part of educating decisionmakers—part of our role as academics.
Weatherall: psychologists can answer questions of perceived similarity; don’t like to talk about confusion: would be more modest in that it would be one piece of evidence that could go into the overall analysis.
Burrell: thinking through what we mean by unrepresentative: are there reasons to think that the ways in which groups differ matter to the perception/confusion question? We too readily assume that they do.
Jennifer Davis: maybe no empirical evidence is ever probative—different approaches include psychologists, polling, neurologists, linguists. They might all have different opinions. No scientific answer to any of these questions. Courts were never dependent on polls/empirical evidence—from 19th c. to now. Rare exceptions, but mostly decisions of judges for both registered and unregistered marks.
Michael Handler: Maybe the most that can be hoped for is to put info in front of judges outside of litigation, in a way that isn’t necessarily intended to impact particular outcomes. Practical impact on bureaucracy/TM offices? Offices may see their roles as faciltators, not gatekeepers: goal is to smooth path to registration. Perhaps they’d respond better to consumer evidence suggesting that confusion is less likely. Work on rules of thumb on similarity may also be more readily accommodated by registration system, for example in manuals of examining procedure, that can then be used in decisions. For example, they do that kind of analysis of surnames in Australia.
Weatherall: There is a strain in the work where particularly Jennifer Berg who does word similarity work has been trying to produce a kind of rating of words. Trying to see whether there are tests that would give us a kind of rule of thumb rating and whether that can be applied in the brand context.
Austin: what do you do with context in this kind of research? Someone who is in a hurry, for example, is just paying less attention. Quality of decisionmaking/perception is a function of emotional state. Payday loans = decisions made under stress.
Burrell: psychologists don’t even want to talk about confusion, so it won’t help answer the ultimate question.
Dinwoodie: CTM context—is there homogenization of European culture that would map onto Davis’ observations about homogenization of imagined British consumer/decline of class differences in imaginaries?
Level of comfort w/acquired distinctiveness surveys—why? Some of the same problems would seem to exist as in confusion cases. Perhaps difference is forward-looking v. backward-looking. Maybe confusion looks more normative as a question than empirical and secondary meaning/acquired distinctiveness looks more empirical, despite normative elements.
Kur: it’s knowledge v. reaction.
Dinwoodie: consumer reacts to sight of bottle by recognizing—why is that different? [Maybe b/c we still have another bite at the apple in constraining rights by defining the scope of rights later on, even if we accept that a sign is a mark by finding secondary meaning.]
Davis: Maybe there’s homogenization in the sense of a decline of a servant class. Marks & Spencers v. Interflora: J. Arnold said he was computer literate and would understand keywords but doesn’t think average consumer would understand. Seems pretty much the same as pre-Directive paternalism. Not a hard & fast rule. Judges are not always well informed.
Dinwoodie: maybe you’d approach a CTM case differently than an EU case.
Davis: it’s helpful for CTM people to think of an average consumer b/c otherwise their task is basically impossible.
Senftleben: You get a second bite at the apple with accepting acquired distinctiveness. Less frightening for a judge to accept acquired distinctiveness by survey; still doesn’t let someone else determine the outcome of the confusion case. [RT: exactly what I was trying to say.]
Dinwoodie: as long as we emphasize to judges that it’s not outcome determinative, why would it be different?
Senflteben: it’s harder perhaps to manipulate the survey: you show a picture and ask if they recognize something. If they can name a brand, that’s secondary meaning.
Mid-point summary: Catherine Ng: Notional consumer: we assume the consumer is rational, reasonably well informed, fluent in 24 languages (Jacob disagrees but Dinwoodie agrees). Unrepresented consumers: those who actually prefer lookalike products/benefit from initial interest confusion. Gaps and convergences between image and real consumers. Substantial portion? Majority? Those questions also continue to arise. And finally the real individual consumer—not necessarily a representative sample. The legislature didn’t necessarily have every single consumer in mind when legislating, but they come to the fore in witness statements, and also when judges say that they can make up their own minds without expert/survey evidence: a sample of one.
UK Gov’t studied desirability of registry: in 1862, law of passing off was found to be very clumsy for traders, esp. those which had to trade internationally in countries w/ a registration system. Hope was to be inexpensive and certain. Trader wishes to know what he can/can’t do.
Martin Senftleben: Though we seem to like the general idea of surveys, we also agree that too much can be bad. Also, this concerns all different areas of TM law: secondary meaning, confusion, and dilution (link established by the consumer, and damage in the form of changed economic behavior, which is almost impossible to show through survey evidence). Maybe the project of finding empirical evidence is a threat to TM law. You can ask empirical questions as a vehicle to find the right questions—a link to reality. We can’t guarantee that survey results will affect the outcome. Solve the perfection problem: if the role of empirical evidence is only to point to the right questions, then they need not be perfect in all circumstances.
Ilanah Simon-Fhima: Challenged TM owners to prove dilution/blurring with empirical evidence; leads to shut down. In the US, Victoria’s Secret and TDRA as a result. TM owners are running scared from the challenge.
RT: As much as I want courts to pay more attention to general psychological/consumer evidence, I find myself in agreement w/Sir Robin that similarity research will not be able to bridge the gap to the legal question of confusion. How similar? Do we look at installed base of cases found confusing in the past, but where the confusion findings were not based on empirics? Confusion examples where there was actual confusion evidence? Context matters too: TRAVATAN/XALATAN had a very high “similarity score” as calculated by the plaintiff’s confusion expert in a case I worked on, versus MEXICAN/AMERICAN, with an even higher confusion score. But even if you were confident about your comparator set, and even if your P and D get a very high similarity score—high enough to place it well within the boundaries of marks that have been found confusing in past cases—the next problem you face is comparing it to the entire universe of signs that might be compared. As it turns out, if there’s any false positives at all in your test, the false positives will so vastly overwhelm the true positives that the test becomes useless. Now there are things you can do about this, like narrowing the set by the fact that people have bothered to litigate about it … but they reduce the utility of bringing in this research in the first place.
Heymann: are we trying to predict harm or prevent it? Shaping might lead us to different policies.
Andrew Griffiths: We’ve been talking about cognitive features, attentiveness—but what about knowledge? Specialist/niche knowledge. We are willing to assume literacy, but what about cultural literacy? Maybe the average consumer is now less likely to regard Homer and Sophocles as conceptually similar and more likely to see Homer and Bart as related. When popular culture becomes niche: does the average consumer know Breaking Bad or not?
Betty’s Kitchen Coronation Street: registrar divided consumers into 3 categories: aware of soap opera Coronation Street; aware of character Betty; and not aware at all. First group would be likely confused; second group less likely to be accused; third group didn’t matter. Because one significant group would be likely to confused, registration rejected.
Dinwoodie: whether similarity is a separate threshold—he’s never thought it could be because similarity has to be related to something. Is an apple similar to an orange? Well, it depends. But can similarity can be refigured so that it is more useful to the TM endeavor?
Weatherall: when psychologists talk about similarity, they’re trying to rate relative similarity—class of goods/services and look how similar marks across that group tend to be—is this mark more or less similar than those marks tend to be from each other.
Ansgar Ohly: PHOSITA: A reflective device partly informed by reality but also by what we want to keep free. Similar in TM, where we grant broader scope to marks with higher degree of distinctiveness. Average consumer is not detached from reality, but also not empirical in the sense that a certain percentage matters. We look at rules that are informed by policy.
Bently: research topic: do judges think they’re making empirical decisions? Even if we have a consensus that they’re making normative determinations.
Davis: there’s no such thing as a normative approach. Things happen and we label them as normative.
Bently: maybe they think they’re making factual determinations and we don’t. Maybe “is this empirical?” isn’t an interesting question, and “what work is the ‘average consumer’ doing?” is a more interesting one. Socialization: when I started doing this work, I couldn’t understand the relation between the legal rules on the page and the conclusions judges were reaching. I had to be socialized into learning what made a close case. Average consumer is part of devices used in socializing lawyers into applying standards we believe are relatively consistent.
Graeme Austin: Take the Champion Spark Plugs case in the US: SCt talks about consumers as interested in the original brand, but also interested in getting cheap used versions. Why is that an empirical question/why frame it that way instead of “we want a vibrant aftermarket”? What are we more comfortable with: trial judges calibrating balance in the aftermarket, or using the rhetoric/traditions of language that help the institutional actors think through the problems. Why is it a problem that we use the consumer to avoid normative questions? Is it because sometimes we don’t like the normative outcomes? (For example, the doctrine that variances, including variance in warranty, results in grey market goods infringing is ridiculous in many cases where the consumer knows quite well what she’s getting; talking about confusion gets us away from the question of what aftermarket we want to have.)
McGeveran: administrative costs matter (as Bob Bone would say). Accuracy versus cost of reaching a decision, it’s pretty easy to tip over the point of overspending on finding the right consumers. The nightmare of market definition: consumer has to be followed, at least implicitly, by “of X.” Consumer of snacks v. chocolate v. this brand can become a pitched battle. Ends up being a fight about knowledge base/assumptions courts can make about them. The more unstandardized you make the consumer, the higher you raise the stakes in that difficult market determination, and that makes him nervous.
Internet user as something distinct from the consumer: he thinks that’s crazy. Understands the judicial impulse to set apart a new technological context, but it’s a different iteration of all the many different experiences consumers have—they are different in different countries, different kinds of stores; bad mistake to create a distinct doctrinal line for this particular situation. As the internet becomes more commonplace, becomes increasingly absurd to think of it as different in kind. If it is justified to think about the internet consumer as categorically different, not clear why there’s not a panoply of different kinds of consumers. (There already is!)
Litman: Internet consumer got split off b/c judges and lawyers found the internet scary and confusing. We got to see something we haven’t seen anywhere: the consumer got smarter, more sophisticated and more savvy very quickly. In ten years we got from easy initial interest confusion to “consumers understand AdWords.” Without discussing the internet specifically, it would have taken much longer for consumers to get smart. Good development in the law has resulted, including for likely confusion.
McGeveran: temporary measure for something new makes more sense; but we continue to have cases treating internet users differently from ordinary users, and cases that don’t respect dynamism.
Litman: it’s only been 15 years; give it time. It’s not consistent intellectually to say we’re only going to make these consumers special, but it helped the corner of the law instead of dragging the rest of the doctrine with it as it almost did in the pre-dilution era involving cybersquatting.
Grynberg: leaves doctrinal detritus of domain names—the rules for them are just different.
RT: There is research showing that consumers are more credulous/less able to make distinctions in new modes of communication. That would support a new medium rule. And we already distinguish between consumers in the checkout line at the grocery store and consumers buying a car.
McGeveran: we use the sophistication/attention of those consumers as one consideration in the multifactor test not the be-all and end-all.
RT: I don’t think so. It’s that the internet cases disproportionately involve comparative advertising, and that’s why you see a generalized “let’s be sensible about this” treatment relatively more than in non-internet cases.
Ng: entire professions dedicated to studying who these consumers are. Marketers have target markets. Is this a way of us as lawyers trying to replicate that upstream knowledge?
Dinwoodie: see in Europe in keyword cases the AG opinion: we just don’t want to adopt a rule that would kill Google’s advertising revenue. This normative opinion may be right, but do we want it everywhere/a rule to protect new technologies?
Davis: So much depends on the assumption of rational processing of information by consumers. But the internet features algorithms that give consumers information they may not be looking for; that’s not something that happens as readily when you go into a physical store.
Kur: internet cases are different in that the dynamic element/speed of consumer adaptation is much more present. Normally if you make a reliable empirical test of how consumers react to a particular offer, then that would still be reliable when the case reaches the Supreme Court, but in internet case by the time it reaches the Supreme Court the behavior of consumers (and of sellers!) may already have changed.
Return to suggestion of asking judges and people in TM offices what they believe they are doing. Germany: judges would say they are making a normative assessment b/c confusion is normative, but they take reality into account so it is a mixed notion. TM examiners would say “we apply the law and go through the factors we are required to consider.”
Simon-Fhima: TM office in UK moved from guardian of register to serving customers—it’s up to other market participants to guard their own interests.
Consumers of pharmaceuticals: either doctors who are very educated, or consumers who are consuming products and care very much about their health. But the application of this idea seems to be less robust: consider how the analysis of similarity goes—any one pharmaceutical is treated as almost identical to any other pharmaceutical, which is at odds with the supposed definition of the average consumer they’re using in the EU.
Grynberg: I always thought of the internet as prompting recall of normative goals of TM in light of empirical situation: a game of analogy in the late 1990s. Is a metatag like a billboard or like stocking goods near branded goods in a store? Davis suggests that the internet is different because you can do a better job of giving consumers a substitute/giving them something they didn’t specifically ask for [RT: though as compared to the impulse buys at the supermarket, I wonder if that’s so], but so what? You can ask whether point of sale confusion/its absence ought to matter.
Stacey Dogan: US law is explicitly normative in its focus on intent. Unfair advantage comes in through the back door: if D is intentionally trying to evoke the P that cuts against it even in absence of intent to confuse. [RT: but only in some cases! Plenty of cases say the opposite!]
Elderly people rely on color/shape of drugs; this is used as justification for copying for generics—we want to make sure that people will take their medicine! You can manipulate judicial instincts to limit scope of TM rights.
Marco Ricolfi: different TM contexts—border measures/special customs provisions exist. It is for customs officials to step in w/out having to exercise judgment if it’s clear there is a violation. That could tell us how normative/empirical a judgment is. (RT: I’m suing the government in the US to find out more about what they seize as counterfeit, which might provide some insights on that.)
Dirk Visser: L’Oreal also considers intent to free ride—he believes intent is a determining factor in EU cases as well, though it’s not officially part of the confusion analysis.