Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu

SRF Interviews Philip D. Harvey

Posted on 26 Sep 2009

For years, Philip D. Harvey has made the cognitive deficits that cause much of the disability associated with schizophrenia a focus of his research. Harvey, a psychologist by training, also founded and directs the Cognition Satellite Meeting, which recently took place in San Diego, California, in conjunction with the International Congress on Schizophrenia Research (ICOSR) (see SRF reports from the morning [SRF related news story] and afternoon sessions [SRF related news story] of the meeting). The Cognition meeting is also a satellite meeting of the Schizophrenia International Research Society and will be convened next spring in Florence.

Inspired by the meeting, SRF asked Harvey to share his thoughts on the field of cognition in schizophrenia with writer Victoria Wilcox. In a conversation that ranged from the field's origins to its future, and controversies in between, Harvey spoke to us from Atlanta, Georgia, where he works as professor of psychiatry and behavioral sciences at the Emory University School of Medicine.

Psychosis: a distinct cognitive process?

SRF: Psychosis and delusions are failures of cognition, aren't they? How do you define what's typically called cognition in schizophrenia research?

PH: That's a very good question. What makes it such a good question is that we understand very well by now what the cognitive impairments, as measured by traditional neuropsychological tests, are in schizophrenia: memory, attention, concentration, etc. But although delusions and hallucinations are cognitive symptoms for certain, we are much less able to identify the cognitive causes of those.

And what's really interesting about that is that delusions and hallucinations don't correlate with real-world functional disability. And so they must be caused by different cognitive processes than those that do correlate with real-world disability—like attention, memory, concentration, and all those things that you heard discussed at our meeting.

SRF: Could that be because delusions and hallucinations don't last as long?

PH: Right; they are episodic. That means the cognitive processes that cause them must be episodic as well.

SRF: That means that they're different processes?

PH: Maybe, because disability and the cognitive impairments that are associated with disability are quite stable. You don't get spontaneous remission of disability like you do of hallucinations. And the other thing that's really interesting is that antipsychotics must improve the cognitive processes that underlie delusions and hallucinations because antipsychotics reduce delusions and hallucinations, but antipsychotics don't improve the cognitive functions that are associated with disability. That's really what's interesting; it's two different domains of cognitive functions.

I could tell you what the hypotheses are about the cognitive processes that underlie delusions and hallucinations, but that's sort of past the point. Let's just say that they're different.

SRF: I would be interested in hearing what those processes are.

PH: We think they're monitoring-type problems. When you're hearing voices, those are things that are happening inside your own brain, so you're confusing something that happens inside your brain with something that's happening out in the world. It should be easy to figure that out, but it turns out that people with schizophrenia who have hallucinations are less able than people who don't have hallucinations to tell the difference between something that they thought and something that they heard, which is quite interesting.

Delusions are probably driven by two processes. Clearly, you have to have some problems in logic that are involved, because if you think about it, some of the things that happen in delusions are completely illogical and couldn't possibly happen. If you ask people with schizophrenia to evaluate the plausibility of other people's delusions, they will say that those are crazy ideas. If you ask them to evaluate the plausibility of their own delusions, they think they're absolutely true. Therefore, it's not a generalized deficit in the ability to understand social plausibility, but rather it appears to be a problem of suspending the rules of rational judgment when it applies to you yourself. That makes sense, doesn't it? In other words, "Of course, it's crazy that the CIA would put a radio transmitter in someone else's head, but they put one in mine. I know it's true, because it's happening to me." And the guy who's really done a lot of important work in that area is Tony David from the Institute of Psychiatry in London, who understands this area very well.

Cognition in schizophrenia: in and out of favor

SRF: Let's go back to the beginning of the evolution of understanding cognition in schizophrenia. Did Kraepelin and Bleuler, or even the people before them, see cognition as impaired?

PH: In fact, Kraepelin described every cognitive impairment that we see in schizophrenia that you can't measure with a computer.

SRF: No kidding.

PH: Absolutely. No new impairments that you measure with a paper and a pencil have really been discovered since Kraepelin described them. And Kraepelin thought that schizophrenia was a frontal lobe disease. Of course, we didn't know as much about the circuitry of memory back 120 years ago as we know now. He thought memory wasn't that impaired, but he didn't mean memory like we mean memory today. He meant long-term memory and rapid forgetting because, don't forget, Kraepelin knew about Alzheimer's disease because Alzheimer was his student.

And Kraepelin's term "dementia praecox" simply implies dementia, cognitive impairment, and functional decline as a central feature of the illness. Then Bleuler thought that cognitive impairment was one of the four fundamental symptoms of schizophrenia, so they both understood that cognitive impairment was present. Where they diverged was their belief about the course of the illness.

SRF: How so?

PH: Well, Kraepelin believed, for the most part, that the illness had a uniformly bad outcome with no possibility of recovery and an at least partially declining course, whereas Bleuler believed in the possibility of recovery. It turns out that they were both right. Kraepelin didn't have the benefit of seeing people treated with antipsychotics, but people who don't respond to antipsychotic treatment look like Kraepelin's patients with schizophrenia; treatment resistance is associated with more profound cognitive functional decline over time. And Bleuler was right, too, because we now know that, even with non-extraordinary interventions, a quarter of people with schizophrenia get better. With targeted interventions, the proportion may be a little bit higher.

SRF: When did the science on cognition in schizophrenia really get its start?

PH: It started in the 1800s. There was a paper published in 1944 by James McVicker Hunt (Hunt, J. McV, Cofer, C. Psychological deficit. In J. McV. Hunt (Ed.). Personality and the behavior disorders. New York: Ronald Press, 1944, pp. 971-1032), who was a famous psychologist, about the first 50 years of research on cognitive impairment in schizophrenia. That's 65 years ago. In fact, I wrote a chapter (Harvey and Neale, 1983) that was published in Progress in Experimental Personality Research as part of the series that Brendan Maher used to edit a long time ago, where we talked about the history of studying cognitive impairment in schizophrenia. There's literature on cognitive impairment that was published in the 1880s.

SRF: In your opinion, what have been the most useful historical developments in the study of cognition in schizophrenia?

PH: I think the most important historical development is the shift that happened; it's a return to an old development. It's that schizophrenia, starting in the 1950s with the advent of antipsychotic medication, became a disease of delusions and hallucinations because of the law of the tool. If all you have is a hammer, everything looks like a nail. If you've got a drug that makes delusions and hallucinations go away, then schizophrenia is delusions and hallucinations.

But all those people who left the hospital didn't go and get a job or a spouse or even a place to live. In the 1990s, we re-realized that cognitive impairment is functionally relevant as we started to make the connection between cognitive functioning and elements of outcome.

Michael Green wrote a paper that was published in 1996 on the role of cognition in disability (Green, 1996), and the data in it were presented for the first time at our cognition conference in April 1995. This paper has been cited over a thousand times and is one of the most important papers on schizophrenia that was ever published. So I think that the group of us, as reflected by the constituency of the cognition meeting, all came to the same realization at the same time.

Deficits worse than in Alzheimer's disease

SRF: What do we know for sure about cognition in schizophrenia and how it differs in schizophrenia versus in people who do not have the disease?

PH: Well, Avi Reichenberg and I have a paper in Schizophrenia Bulletin (Reichenberg et al., 2009), where we took Evelyn Bromet's data from the Suffolk County High Risk Project and compared post-first-episode people with schizophrenia, bipolar disorder, and unipolar depression, all of whom had been termed psychotic at the time of their first episode. We found that the differences in cognitive function were only a question of degree and not profile. The profiles were identical; the severity is what varied.

Now, cognitive impairment in schizophrenia is definitely differentiable from Alzheimer's disease. In the work we did for 15 years studying old people at Mt. Sinai in New York, we were clearly able to show that there's a difference in both profile and course between Alzheimer's disease and schizophrenia. People with Alzheimer's disease have more rapid forgetting, but all other aspects of cognitive functioning, when you match subjects for age and global cognitive impairment, are actually more impaired in schizophrenia than in Alzheimer's disease. And so when psychiatrists say to me, "Cognitive impairment in schizophrenia is really subtle; it's not like Alzheimer's disease," I say, "No, it's not; it's much worse." And we published that paper in 1996 (Davidson et al., 1996), so that's been out there for a while.

SRF: How does schizophrenia compare with bipolar disorder in that regard?

PH: People with bipolar disorder are, relative to healthy people, about half as impaired as people with schizophrenia. However, people with bipolar disorder may start out a little better before their illness begins, because people with bipolar disorder on average probably have an IQ that's a little above average, and people who are going to develop schizophrenia are probably below average expectations. Thus, people with bipolar illness may have the same risk for worsening with illness as people with schizophrenia.

SRF: How common are cognitive deficits in schizophrenia? Do we know what percentage of people with schizophrenia has cognitive deficits?

PH: There was a paper published in the journal Neuropsychology in 2005 with the title, "No, it is not possible to be schizophrenic yet neuropsychologically normal" (Wilk et al., 2005). The argument is that everyone with schizophrenia has some element of cognitive change relative to what they would have been like if the illness had never struck.

SRF: There is an emerging body of work that supports the idea of very subtle deficits in all areas of the brain in schizophrenia, and your article with Dwight Dickinson (Dickinson and Harvey, 2009) argues that a generalized cognitive deficit lies at the heart of the disease. Could you sketch out that idea and tell us what reaction you've had to it?

PH: The idea is that there are multiple cognitive impairments, that most systems are affected, and that there's probably no real focal impairment. The reason for that is also partially a psychometric argument. It's a boring argument, and I can make it for you in one sentence. Basically, discrepancies between ability areas in an individual have to be quite large in order for a neuropsychologist to consider the difference to be clinically significant. If you're impaired on everything a little bit, It's really not even possible for the tests to detect a substantial impairment because they're not reliable when you start to get out to three standard deviations away from the mean.

And studies that have looked at focal deficits and argued that schizophrenia is a temporal lobe disease or a frontal lobe disease have often only measured frontal lobe or temporal lobe functioning and have not adjusted for performance on global indices. So I don't think there's much of anyone saying that schizophrenia as a broad entity is caused by a dysfunction in a single system.

In fact, some of the really interesting work done by Danny Weinberger and colleagues (Meyer-Lindenberg et al., 2005) at the National Institute of Mental Health, as well as by other researchers (Ross and Pearlson, 1996), has suggested that there are a set of connected cortical and subcortical sites in a network that are impaired in schizophrenia, which would then easily produce this global deficit. In other words, if your anterior frontal cortex and your superior temporal gyrus and your inferior parietal lobe and your hippocampus and your corpus striatum are all linked in a network, and there's a failure at any point in the network, everything is going to be impaired. You blow one fuse, and all the lights get dim.

SRF: How do you even know if there's a real difference between domains and that one area of functioning is much more impaired than another?

PH: There are some really exciting neuroscience developments that you saw in our morning session (see SRF conference story), where we pulled a lot of people together. Using imaging and highly sophisticated, yet very simple, tasks, it turns out to be possible to elicit regional changes in brain functioning. Whether or not those regional changes are important above and beyond the global signal is truly an open question.

A pair of paradigms

SRF: Are there dominant schools of thought about the nature of cognitive deficits and their causes in schizophrenia?

PH: I think the schools of thought are more aimed at the approach you should use to investigate; you saw that at our cognition meeting, right? In the morning, it was the neuroscience approach, and, in the afternoon, it was sort of the integrated transitional approach, where we use tests that everyone understands, measure functioning, and try to use research designs, imaging techniques, and statistical approaches to basically understand the course and correlates of changes in important cognitive domains (see SRF conference story).

The reason that people still use traditional neuropsychological tests is because they have been developed over a century to be sensitive to disability. And those of us who are primarily interested in disability and functioning don't yet know if sophisticated neuroscience tests have anything to do with real-world outcomes.

Remember, we started our conversation talking about how delusions and hallucinations don't have anything to do with real-world outcomes in most cases. Imagine that those sophisticated cognitive neuroscience tests helped us to elegantly understand the underpinnings of delusions and hallucinations. That would be great but, at the same time, I think that it would not help us understand disability. The schools of thought are, one, refine the measurement of brain dysfunction, and two, have a clinical anchor point and see how things relate to it.

Known correlates, unknown causes

SRF: Do we know anything for sure about what causes cognitive deficits in schizophrenia?

PH: There's a lot of debate about that, and I don't believe that we really know what the cause is. We know what the correlates are; it's dysfunction in this linked network. It probably has something to do with some basic neuropathological processes; there are a number of findings of changes in certain cells, like David Lewis's work looking at the GABA system and changes in chandelier cells. The glutamatergic system, of course, is influenced markedly by the dopamine system; they interact with each other quite closely.

There are several levels at which there could be a cause of cognitive impairment. It could be brain structure, it could be brain function, or it could be brain neurochemistry. We've looked a lot at neurochemistry because antipsychotics work neurochemically.

SRF: That's an example of the tools....

PH: Exactly. People want to impact cognitive functioning in schizophrenia, so they develop drugs that manipulate other transmitters. There aren't any drugs that we know that manipulate brain structure yet, although there are certainly drugs in development that are aimed at treating white matter dysfunction and spinal cord lesions, for example. That will be an alternate approach, an approach aimed at changing the structure in the brain. Then, of course, certain drugs like amphetamine work on multiple transmitter systems and increase the overall activity in the brain. Regional activation is changed with certain drugs across more than one neurotransmitter system.

We have models for looking at the cause of cognitive impairment, but we don't have any answers yet. At least we're thinking clearly about that.

Controversies and questions

SRF: What are the biggest controversies and lingering questions about cognition in schizophrenia?

PH: One of the lingering questions is that of the absolute functional relevance of cognitive impairment, because it's correlated with aspects of outcome. It may be the best single correlate of functional disability, but is it a good correlate?

Also, how much do we need to understand the societal backdrop before we can really understand cognition? Another controversy concerns which method is the right way forward: Is the neuroscience method the right way forward? I think that it has a lot of strengths, and that's why we put together such a big neuroscience perspective at our cognition meeting.

But it's not clear yet whether these neuroscience tests will actually spur drug development that is aimed at the distal outcome of disability. I'm working with some of the people who are involved in CNTRICS—I'm a consultant to a couple of their studies—and we are putting some of these functional measures in there to see how these things work.

SRF: Do you think the neuroscience methods and the neuropsychological approach will ever be resolved so that they can be combined in some way?

PH: It's interesting that you ask that. We wrote a paper on that in 1994 that was published in Schizophrenia Bulletin (Serper and Harvey, 1994), and basically someone wrote a commentary (Strauss and Summerfelt, 1994) that said neuropsychological tests can never be integrated with neuroscience measures because they're just old and no good. I think that is wrong. We are still working on whether we can integrate these perspectives 15 years later.

I think it would be great if we could identify neuroscience measures that would completely account for all of the variance in prediction of the disability outcome that is accounted for by clinical neuropsychological tests because then we could use that for treatment development. But I don't think we're ever going to have clinicians in practice putting 128 electrodes on someone's head to assess him or her in the office.

SRF: I'm intrigued by your comment about the influence of society on the disability that results from the cognitive deficit.

PH: In a study that was just published in the American Journal of Psychiatry (Harvey et al., 2009; see SRF related news story), we collected the same cognitive and functional assessments on people with schizophrenia who lived in the Bronx and who lived in a rural county in Sweden where there are more cows than people. And what we found was that on the UPSA (University of California, San Diego, Performance-Based Skills Assessment), the functional outcome scale I used in my study on cognitive decline in old people with schizophrenia, scores in the rural county in Sweden and in the Bronx were absolutely identical.

And clinician ratings of people's ability to function in the real world were absolutely identical. But in Sweden, about 80 percent of the patients were living independently in the community, and in the Bronx, it was 35 percent. And the reason for that is, in this rural county in Sweden, they give you $1,000 a month in disability compensation if you have schizophrenia, and they pay 97 percent of your rent. The only skill you have to have to live independently is to be able to pay your landlord. That's where society comes in. In other words, in Sweden, there's absolutely no correlation between your ability and your outcome.

SRF: So it's like other types of disability in that it's shaped by society?

PH: Yes; why would you try to work if it was hard to find a job and someone was going to pay your rent anyway? And why wouldn't you live independently if someone took care of everything for you? So the disability these patients experience is not expressed in terms of their real-world outcomes, but in terms of their potential. Their potential is the same; their outcome is different. If we decided in the U.S. that we were just going to give $4,000 a month to everyone who had a serious mental illness, everyone could live independently.

SRF: I'm sure you've heard arguments that researchers could learn from the study of people who are happy and well-adjusted or aging well instead of just focusing on people who are ill or impaired, as in the field of positive psychology or the study of successful aging.

PH: That's absolutely true. In fact, when I was at Mt. Sinai, we had a big grant from the MacArthur Foundation to study healthy aging, and we published papers on the brain functioning of older people who were very high functioning. What we found then (Harvey et al., 2009; Hazlett et al., 1998) was that healthy older people were able to recruit additional brain regions that healthy younger people don't need to maintain adequate performance on memory tests. For example, a memory test that is moderately challenging to someone who is 40 years old and high functioning would be impossible for the average 80-year-old, because your memory declines naturally with aging, but these 80-year-olds who were still high functioning—most of them had doctoral degrees and were still working full time—were able to use parts of their brain that the healthy younger people didn't need to use to keep their performance the same. What we learned subsequently to that, in studies of people with schizophrenia spectrum conditions (Koenigsberg et al., 2005), is that people with these conditions were unable to use parts of their brain that would be adaptively useful to perform normally, so the idea of being able to allocate brain effort is a trait that ranges from impaired in people with schizophrenia to absolutely outstanding in healthy older individuals. We learned a lot from that.

SRF: Interesting. Some people fault DSM-IV for not emphasizing cognitive impairment enough as a key feature of schizophrenia. Should DSM-V correct that?

PH: I was just on a symposium at the World Psychiatric Association (WPA) the week after the ICOSR on the ins and outs of adding cognitive impairment to the diagnostic criteria. And the only downside is if you make it part of the criteria, then you're essentially putting an assessment demand on the diagnosticians, and they may inaccurately diagnose the patient simply because they have no way to assess them correctly. As it is, we have diagnostic criteria that generate a group of people, 100 percent of whom have cognitive impairment. Why would you need to put it in the diagnostic criteria? The current diagnostic criteria capture it.

DSM-V probably should include language saying that schizophrenia is notable for the presence of cognitive impairment in essentially every individual who has it, but I can tell you why everyone who has schizophrenia has cognitive impairment. It's very simple. It's Criterion B.

SRF: Which one is Criterion B?

PH: That is disability in functioning, a decline from premorbid levels that lasts 6 months. If you require disability, the cause of disability is cognitive impairment.

SRF: So, indirectly, it's in there.

PH: Absolutely, it is. It's just that the language needs to be in there because what we need to do is have clinicians know, when a cognitive-enhancing treatment finally does make it into prime time, that people with schizophrenia should receive it because it will reduce their disability.

Cognition's spillover to other areas

SRF: Are there implications of cognitive deficits in schizophrenia for research in other areas of schizophrenia? In other words, should people who are studying other areas be interested in cognitive deficits?

PH: Absolutely. Here's an example: Recently, there's been a lot of buzz about using CBT to reduce persistent psychotic symptoms in people who are treatment nonresponsive.

SRF: CBT—that's cognitive behavioral therapy?

PH: Exactly. And some meta-analytic reviews have suggested that CBT on average doesn't work that well, although it seems so promising. One of the reasons it might not work that well is that CBT is a learning procedure and that people with treatment-resistant psychotic symptoms have the worst cognitive impairments of people with schizophrenia. You're preselecting a population that has a hard time learning, and then you're trying to teach them.

SRF: I see the problem.

PH: If you would screen for a cognitive impairment and adjust your results accordingly, you'd probably find that these interventions work quite well. It's just that they don't work quite as well in people who are more cognitively impaired.

SRF: But then does that get back to the training issue in that people don't know how to test for these impairments?

PH: Yes, it's too bad, because you saw our article with Dwight Dickinson (Dickinson and Harvey, 2009). Basically, the most informative test takes 90 seconds to administer.

SRF: So that could be a fairly simple thing to remedy.

PH: Yes, if you're looking for just a rough global estimate. At WPA, I talked about how hard would it really be to do a cognitive assessment on people with schizophrenia: not very hard if you are willing to take a very general approach to understanding what you are interested in.

SRF: Do cognitive deficits in schizophrenia have any implications for existing theories of the disease?

PH: I think that most theories of schizophrenia now are brain-based theories, and cognition originates from the brain. I think that most comprehensive theories are at the very least touching on cognitive impairment, if not addressing it directly.

Study strategies

SRF: What are some of the unresolved measurement issues in this area? For instance, the paper by Dickinson and Gold in Schizophrenia Bulletin last year (Dickinson and Gold, 2008) talks about overlap among neuropsychological measures.

PH: That's why if you just use one neuropsychological test, that doesn't mean that you wouldn't find impairments on something else that you might administer. The reason for that is really simple: Neuropsychological tests were designed to be screening measures to capture all kinds of impairment. A good neuropsychological task is sensitive to impairment, no matter where it originates.

SRF: Does that mean that you cannot use those measures for distinguishing domains?

PH: Our analysis of the CATIE (Clinical Antipsychotic Trials of Intervention Effectiveness) cognitive data suggests that the best-fitting factor model for cognitive impairment, measured using standard neuropsychological tests, is one factor. And that was with a sample size of 1,300 patients, so we had the most power to find separate factors and to do our tests. I think that's why standard neuropsychological tests may be good outcome measures in clinical trials, but they're certainly not going to be measures that you want to use to guide drug development because you don't know exactly what you're measuring. I think that's Jim Gold and Dwight Dickinson's point.

SRF: You can't use them to untangle the various cognitive domains because they're too intercorrelated?

PH: Yes; they're too related to each other. And if you think about the contributions to something like processing speed, I mean, if you're really slow, then every test turns into a speeded test. In a test like digit span to measure working memory, you're supposed to read one digit every second. That's not viewed as a processing-speed test; it's viewed as a working-memory test, but if you're three times as slow as the average healthy person, then that becomes a speeded test, because one digit every second is too fast for you.

SRF: Is there any hope for people who want to study a specific domain? How would they measure it?

PH: Sure; that's where the neuroscience stuff comes in. The domains are labeled: for example, verbal episodic memory. Episodic memory is a concept that's very fractionable, right? You can look at the ability to learn words. You can look at the ability to learn faces, you can look at the ability to learn objects, and then you can look at interference processes. There are a lot of ways that these things could be fractionated and studied in more detail. Lots of neuroscience people do that already, but you wouldn't want to use the Hopkins Verbal Learning Test as your only measure of memory.

SRF: Are there other unresolved methodological issues or blunders that you see being made?

PH: Oh, I don't think that there are a lot of blunders going on now. Every now and then, I'll review papers that are just silly. I think that the biggest potential problem, and this is something that's resolvable—for example, this is what CNTRICS is trying to resolve—is that many people's highly sophisticated neuroscience tests are not exactly the same as someone else's. It's almost the opposite problem of what you see with neuropsychological tests, which are highly correlated with each other.

If you change certain of the processing demands in these more sophisticated measures, the test could turn out to be very different. For example, different versions of the Continuous Performance Test correlate with each other very poorly; when you change the processing demands subtly, you might be measuring something completely different. So the lack of standardization to date in neuroscience measurement could possibly obscure some of the findings. It's not a blunder, and that's what CNTRICS (Carter and Barch, 2007) is trying to remedy, to see if when you standardize these tests, they're actually useful in multisite trials.

SRF: How useful are animal models for studying cognition in schizophrenia?

PH: The biggest problem with animal models is that there aren't any animals that get schizophrenia.

SRF: That's a big problem.

PH: Yes, and that's the origin of Tim Crow's theory (Crow, 2000) about schizophrenia being a uniquely human disease, and cortical lateralization and language being the path to schizophrenia. The logic makes sense because, when you try to induce a schizophrenia-like state in animals, you have to do it neuropharmacologically or with a lesion, which means that if you're wrong about the cause, then you're wrong about the model. It's probably easier to have animal models of things like problem solving, working and episodic memory, than of schizophrenia. It's probably easier to model the cognitive impairments in schizophrenia than to model schizophrenia.

SRF: That makes sense.

A rocky road to remedies

SRF: What's the state of the art in treating cognitive deficits in schizophrenia? Are there really any good treatments?

PH: There are two paths. There's the pharmacological path, and there's the cognitive remediation path. Both were discussed in detail at the ICOSR. There are some promising results from cognitive remediation studies, and there have been few promising results from pharmacological studies.

SRF: Why do you think there's been so little success in the pharmacological studies?

PH: I think one of the main reasons is that we don't know how antipsychotic medications interact with the various potential cognitive-enhancing compounds, unfortunately. It may very well be that the pharmacological processes that we are trying to manipulate may be downregulated in some way through the use of antipsychotic medication.

SRF: Oh, I see, because the study population is taking antipsychotic drugs.

PH: Yes, we've got a population of people who are taking drugs that blockade both the dopamine D2 receptor and the serotonin 2A receptor. Serotonin regulates a variety of other neurotransmitter functions, so if you down-regulate serotonin, you may be downregulating other neurotransmitters at the same time. Conversely, if you are down-regulating dopamine D2, that may kick off some kind of cascade that has a stronger signal than a pharmacological agent that's aimed at manipulating some other transmitter. This is just a theory. But what happens is—I just recently saw some data, although it was at an advisory board so I can't tell you where it came from—that showed that a cholinergic M1 antagonist has differential effects depending on what antipsychotic the animal was being administered. In other words, it may very well be that different antipsychotics have different adverse impacts on some of these neurochemical probes that are being used.

SRF: And antipsychotic drugs may not be the only drug that these people are taking.

PH: Absolutely. But what happens is that antipsychotic drugs are what all of these people are taking. They may be taking other drugs, but they're all taking antipsychotics, or they couldn't get into the studies.

SRF: Could you give us a sense of what looks promising for the future?

PH: Well, there's a lot of promise. There are several levels at which cognitive-enhancing drugs impact. We don't know what the dose is; drugs are very hard to dose in the abstract. You don't know if you'll need more or less of the drug if you administer it with an antipsychotic compared to by itself. And we also don't know whether some of these transmitter systems are as implicated in cognitive impairment in schizophrenia as they are in animals. People are trying to push forward on this, and I think it's a very important thing.

Early warnings and opportunities?

SRF: What kind of cognitive deficits do you see in the premorbid stage in people at higher risk of developing schizophrenia? And what can you learn from those?

PH: It's interesting because people in the premorbid stage do show some of the same impairments as people who have the disease. Although people with schizophrenia do have detectable cognitive impairments when they're in the premorbid state, the level of impairment is subtle enough that it's not really a diagnostic indicator. But they show the same type of cognitive impairments: they show impairments in episodic memory and attention. They also may show impairments in executive functioning, although that's slightly more controversial.

SRF: Some researchers at ICOSR suggested identifying premorbid deficits in people who are at high risk of developing schizophrenia and then targeting those people for intervention regardless of whether or not they develop schizophrenia (see SRF related news story; see SRF related news story). What do you think about that?

PH: Well, the only shortcoming of that is that if you insist that people have to have very substantial deficits, you'll miss quite a few people who will develop schizophrenia. If you say that you're going to target everybody who is half a standard deviation below the healthy mean, that means you're going to be intervening on 17 people for every one who would get schizophrenia because that cutoff is like the 17th percentile.

SRF: What about just tackling the cognitive deficits without trying to prevent schizophrenia?

PH: The thing is that the impairments that you see are within the normal range during the prodromal period. They're just low average. What happens is that more people who develop schizophrenia are performing at the 17th percentile than people in the healthy population are. But most people who are performing at the 17th percentile are in the healthy population and are not going to develop schizophrenia.

Outlook for the future

SRF: Recent genetic studies suggest that hundreds, maybe even thousands, of genes, affect schizophrenia risk, each with very small effect. You mentioned Dan Weinberger earlier, and he said that genes don't code for disease, or even cognition, so is there really any hope that genetic research can have an impact on cognition research or treatment?

PH: Combinations of genes do code for cognition, and the processes that genes influence affect cognition. The perfect example is the COMT polymorphism that Weinberger and his group have studied in detail. It accounts for variance in cognition in the healthy population and in people with schizophrenia, although it's most likely not a susceptibility gene.

SRF: Are there areas related to cognition in schizophrenia that you feel should get more research attention and funding? Are there maybe other areas that you feel have reached the end of their useful life?

PH: I'm not really in a position to comment on what research shouldn't be done. I do think that the area of cognition and functioning in schizophrenia is valued quite highly by the current group at the National Institute of Mental Health, and that's a good thing. In the past, it might not have been valued as much.

From lexical networks to aging with schizophrenia

SRF: I'd like to talk to you about your own background. You're a clinical psychologist?

PH: Yes.

SRF: And you see patients?

PH: Yes. I see lots of research patients, and I see occasional patients for psychological assessment consults.

SRF: How did you get interested in the whole area of cognition in schizophrenia?

PH: I realized that cognitive impairment was substantial in schizophrenia and so I sought out a graduate school mentor, John Neale, who is a well-known psychologist at SUNY (State University of New York) at Stony Brook who was studying cognition in schizophrenia.

SRF: How did you come to realize that it was important?

PH: When I was an undergraduate, I was interested in cognitive science even though they didn't call it that then. I did an undergraduate honors research project on activation models of lexical networks in healthy people, and was interested in the brain and how it functioned.

SRF: You've done some interesting work on older people with schizophrenia. I'm wondering if you would talk a little about that.

PH: We were interested in the fact that, although people have argued that cognitive impairment in schizophrenia is stable over the lifetime, when we went to hospitals and saw very old patients, they showed impairments that absolutely could not have been their lifetime level of functioning. We started trying to identify who it was that showed these declines and started to look at the course of cognitive functioning in these older patients. We were interested in the intersection between symptoms and cognition, and we managed to identify some risk factors for extraordinary cognitive decline in these older patients.

We worked forward from there. We had a very large sample of patients. I have a slide from a grand rounds talk that I used to give that said that we had about 400 patients in our sample, say, whose Mini-Mental scores were less than 10, and 50 of them were college graduates. I used to joke that the City University of New York has open admissions, so if your Mini-Mental was 6, conceivably you could be admitted to the university. But they don't have open graduation, so it's very unlikely that you could be a college graduate if your Mini-Mental had been 6 your whole life. They had to have declined.

And then the question is: Did they decline more than you'd expect with healthy aging? Most people who age and are healthy don't wind up with Mini-Mentals of 6. It's a fairly subtle thing. We've found that neuroscience tests are more sensitive to decline in our patients than standard neuropsychological tests are. So it might very well be that the reason that you don't find cognitive decline over the lifespan in people with schizophrenia is that the tests are inadequate to detect it.

SRF: Do you feel that aging with schizophrenia is a popular area of research, or do you feel that it deserves more research attention than it's received?

PH: You know, for a while, there were two of us studying aging in schizophrenia; now there's one—just Dilip Jeste basically, because I haven't found the old patients around here in Atlanta. There's very little research done on this. And, in fact, you'll see people who follow up first-episode patients for a year and find no cognitive decline, and they'll say that that proves that cognitive impairment is stable over the lifespan in schizophrenia. If you followed 23-year-olds until they were 24 and tried to make broad-ranging conclusions about the course of cognition in the healthy aging population, you'd look pretty dumb. You'd conclude that Alzheimer's disease is a myth, that dementia is invented by the neurologists for their business, and that no one ever dies.

SRF: Why do you think there's been so little research attention on aging and schizophrenia?

PH: Because, first of all, schizophrenia has a reduced life expectancy in the community. People with schizophrenia, when they get older, since they've been disabled for their whole lives, are more likely to wash up in low-end nursing home care. You don't know where to find them.

SRF: It's difficult to do the research.

PH: Yes, plus when the nursing home operators get the person with schizophrenia into their nursing home, they change their diagnosis.

SRF: They change their diagnosis?

PH: Yes, they don't carry forward their schizophrenia diagnosis; they diagnose them as having COPD (chronic obstructive pulmonary disease) or whatever. So if you do a Medicare diagnosis search, you can come to the embarrassing conclusion that a prominent researcher at Rutgers came to when he said that there were only 3,000 people with schizophrenia living in nursing homes in the entire country. And we had found 10,000 of them in New York City alone because we tracked them from the hospital into the nursing home. We didn't try to look them up using Medicare information.

SRF: I would have thought they would diagnose them with dementia.

PH: That, too. If you diagnose them with having Alzheimer's disease, and you don't carry forward the schizophrenia diagnosis, it looks like they have dementia.

A foothold for progress

SRF: In 1994, you coauthored a book with Richard Keefe called Understanding Schizophrenia, A Guide to the New Research on Causes and Treatment (Keefe and Harvey, 1994). The first chapter was called "The Reason for Hope." Are there more reasons for hope today?

PH: I think that in the 15 years since we published that book, there's been a tremendous set of advances, particularly in genetics, antipsychotic treatment, and also in the understanding of the importance of cognitive functioning, even though we talked about that in our book way back then. I think that the groundwork has been laid. The reason we said there's reason for hope is because people knew in 1994 that schizophrenia is not caused by bad parenting or by adverse social environments.

I think that there's reason for hope because we've got a much better groundwork for understanding the neurobiology and brain functioning in the illness, and we're focused on the stuff that's important. As to whether or not those developments have been reduced to practice, not quite yet.

SRF: Is there anything that I haven't asked you about that you feel is important to talk about regarding this whole area?

PH: I think you've covered the breadth of it; you're obviously very well-prepared. The only other thing is the idea of developing measures that directly assess disability. You saw a lot of that stuff being presented at our conference in San Diego—all of the direct disability measurement stuff with UPSA, for example.

SRF: Thank you. You've been great about answering all of our questions, and I really appreciate your talking to me.

PH: You're very welcome.