There really could be more long-term qualitative studies exploring the nuances of these experiences of improvement and easing of depression, including further shifts over time. In addition to the diversity of responses to antidepressants, something you didn't discuss here is diverse trajectories of sensitivity to relapse or trailing-off of benefits, either when remaining on the drug or stopping it. The effects of cycling on and off periodically, or possibly developing a tolerance to earlier therapeutic effects, probably has light to shed as well on how antidepressants work initially.
My own anecdotal experience is that the first time I ever took Prozac it helped dramatically (if perhaps not magically), a classic positive response. But most of that wore off over the year. In the ensuing decades, each time I stopped and restarted SSRIs (even at the highest dose or switching to a different SSRI), each successive time the response was less and wore off more easily. But what I really noticed is my mood grew less resilient and emotions more blunted, whether on or off them. I would never generalize from this and I'm not anti-medication, but I do wonder about SSRI's habituating effect and how that bears on what you discuss above.
Thank you Chris. You bring up very important points and I agree that considerations of course and tolerance-like effects are crucial. You are definitely not alone in your experiences
Thank you for the nuanced discussion, Awais, and for the call out to the review I co-authored with Dr. George Dawson. I hope readers will take a look at the full article in Psychiatric Times. Granted that there are several plausible mechanisms of action for antidepressants, I believe that the more objective the measure of emotional "blunting", the less convincing the evidence for blunting as an explanation for antidepressant efficacy. For example, we cited studies using the Oxford Questionnaire on the Emotional Side-effects of Antidepressants (OQESA or OQuESA)—arguably the “gold standard” of assessment in this area. (Again: this is not to discount the importance of assessing blunting or recognizing that it can be quite upsetting to patients). It is also worth noting that the term "antidepressant" needs a bit of parsing. To my knowledge, emotional blunting is virtually never seen with bupropion, which probably enhances dopaminergic and/or noradrenergic function, and is arguably an under-utilized agent for depression.
Anecdotally, when I’ve stopped, started, or changed doses of my SSRI I've noticed changes in cognition within just a couple of days, which precede the mood changes. One is in depressive nihilism, the tendency to think that nothing matters because we're all going to cease existing, which I get (minimally) every winter. This is interesting to me in light of the global reduction in significance some people report. The SSRI I take enables me to care about things *besides* the fact that everybody is going to die, perhaps by making that particular fact less significant to me. Maybe what manifests as increased indifference for some is necessary for others to disentangle knots in their landscape of mattering that are distorting their sensitivity to everything else, by giving a little slack. Some people might need that slack to keep more knots from accumulating, others might only need to shake out one.
The other domain is social cognition. I'm intrigued by the idea that SSRIs might directly or indirectly impact something like an internal status thermostat. A reduction in negativity bias could accomplish the same thing, though (and while that's a super general mechanism, at least it's an actual mechanism--I always felt neurotransmitter hypotheses were pseudo-explanatory, insofar as the relationship between the neurotransmitter and the pathological state is just kinda stipulated).
I think many bipolar people (and also some with MDD) have what you could call chronic inter-episode dysthymia--persistent issues with motivation, energy, and the other variables that rise and fall with the dramatic mood episodes--that doesn't always improve when mood episodes are under control. SSRIs just don't seem to help much with this, and, again anecdotally, SNRIs and bupropion seem to work better for acute-unsticking-from-a-stuck-place than for chronic-resetting-to-a-more-functional-default-place.
I wonder how the two contender hypotheses can be reconciled with the following findings:
Re shifts in emotional processing:
The line of research on acute or sub-acute effects of SSRIs on facial expression recognition or emotional processing produced quite inconsistent findings, complicating any interpretation as possible mediator. A prediction algorithm for antidepressant response using these markers of emotional processing barely was able to predict antidepressant response better than chance, even while taking improvement in symptoms within the first week into account (see https://www.nature.com/articles/s41386-021-00981-z and its supplement).
Re cognitive flexibility and neural plasticity:
Since stressful living conditions are associated with higher levels of depression how would the creation of a "window of plasticity" be desirable in the first place? Would it not simply amplify the toll of the maladaptive environment? Also, I would assume that introducing a "window of change" would widen the range of outcomes under antidepressant treatment compared to placebo, however I am not aware of any statistical manifestations of this increased variance. Rather, the variance between antidepressant and placebo arms in RCTs seems to be identical (see e.g. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0241497)
The study by Tang et al. I find very interesting, I did however not find much other literature replicating or refuting the results. I wonder though how an increase in neuroticism and the accompanying buffering against stressful conditions aligns with a proposed "window of change"
The poor performance of markers of emotional processing doesn't reflect well, I agree. I don't know if that is enough for us to say that this hypothesis is likely false. Research on effects of serotonin (e.g. recent study which investigated direct release of serotonin) does seem to suggest a role for emotional processing.
Cognitively flexibility... it's still not clear whether this should necessarily translate into a wider range of outcomes under antidepressant treatment. Here's what Page et al. 2024 say about it:
"The neuroplastic benefits of antidepressants are likely enhanced when they occur in the context of positive experiences in the patient’s everyday environment. People living in more favorable conditions (e.g., employed, college educated, high income) may be more likely to benefit from an antidepressant alone than those not living in favorable conditions [91]. Similarly, psychotherapy, which provides a supportive environment to facilitate positive changes in thinking, feeling, and behaving, enhances the therapeutic potential of antidepressants [92]. In this way, antidepressants can be thought of as opening a window of plasticity that allows the benefits of a positive environment to take hold in neural circuits. By contrast, if experiences in the environment are negative, rodent studies suggest that antidepressant treatment may even be detrimental because enhanced neuroplasticity in an adverse situation facilitates learning that the environment is harmful [93, 94]. However, not all findings support the notion that antidepressants simply make people more susceptible to their environments: rodent studies also show that antidepressant administration prior to stress exposure can buffer against the effects of adversity [95, 96]. Therefore, antidepressants may confer resilience in the face of negative experiences in addition to their ability to enhance the benefits of a positive environment." (https://www.nature.com/articles/s41380-024-02625-2 )
The variance literature is interesting. I don't understand the methodology well enough to know whether the variance issue is considered settled. Some researchers I asked a while back say they weren't convinced yet that variance is identical. And the Stone et al. 2022 (https://www.bmj.com/content/378/bmj-2021-067606 ) showed a greater likelihood of large response in antidepressants vs placebo, which sounds as if it should manifest as increased variance, so I am not sure why it doesn't show up that way.
Yeah, as far as I know, no one has attempted a replication of the neuroticism trial. Would be great if they did!
What are your thoughts on how antidepressant work?
I really appreciate this post, both in terms of your analysis and synthesis and the sources and links. Following up on your last paragraph, any thoughts on what to do with the term "anti-depressants"? Is there a term that does justice to the possible treatment mechanisms or symptoms/traits treated? (a term for communicating with patients and the lay public). Or given the complexity, is it better to avoid such a term? Also, is there a one-sentence explanation of how serotonergic drugs work that you provide to your patients? Or do you describe several possible accounts?
Thanks Thomas! Since antidepressant is the more familiar term to many in the public, I often use it. I try to convey two things: one, they don’t fix any “chemical imbalance,” and are just helping manage symptoms better; two, they are thought to create positive shifts in how brain processes emotions but the actual mechanisms are more complicated and still being worked out. For folks interested in more, I go into further details.
Really interesting update/summary, thank you. I'm intrigued by the paroxetine-neuroticism paper and would be interested in what a statistician (or other people) would think about the "controlling" for neuroticism. I looked up the NEO questions and they are things like "I am seldom sad or depressed" and "I rarely feel lonely or blue." When the two sets of questions are so similar, how can we interpret this? The paper doesn't investigate individual item scores on the HAM-D or NEO (sample size too small perhaps) but that might be interesting.
There really could be more long-term qualitative studies exploring the nuances of these experiences of improvement and easing of depression, including further shifts over time. In addition to the diversity of responses to antidepressants, something you didn't discuss here is diverse trajectories of sensitivity to relapse or trailing-off of benefits, either when remaining on the drug or stopping it. The effects of cycling on and off periodically, or possibly developing a tolerance to earlier therapeutic effects, probably has light to shed as well on how antidepressants work initially.
My own anecdotal experience is that the first time I ever took Prozac it helped dramatically (if perhaps not magically), a classic positive response. But most of that wore off over the year. In the ensuing decades, each time I stopped and restarted SSRIs (even at the highest dose or switching to a different SSRI), each successive time the response was less and wore off more easily. But what I really noticed is my mood grew less resilient and emotions more blunted, whether on or off them. I would never generalize from this and I'm not anti-medication, but I do wonder about SSRI's habituating effect and how that bears on what you discuss above.
Thank you Chris. You bring up very important points and I agree that considerations of course and tolerance-like effects are crucial. You are definitely not alone in your experiences
Thank you for the nuanced discussion, Awais, and for the call out to the review I co-authored with Dr. George Dawson. I hope readers will take a look at the full article in Psychiatric Times. Granted that there are several plausible mechanisms of action for antidepressants, I believe that the more objective the measure of emotional "blunting", the less convincing the evidence for blunting as an explanation for antidepressant efficacy. For example, we cited studies using the Oxford Questionnaire on the Emotional Side-effects of Antidepressants (OQESA or OQuESA)—arguably the “gold standard” of assessment in this area. (Again: this is not to discount the importance of assessing blunting or recognizing that it can be quite upsetting to patients). It is also worth noting that the term "antidepressant" needs a bit of parsing. To my knowledge, emotional blunting is virtually never seen with bupropion, which probably enhances dopaminergic and/or noradrenergic function, and is arguably an under-utilized agent for depression.
Best regards,
Ron
[Ronald W. Pies, MD]
Anecdotally, when I’ve stopped, started, or changed doses of my SSRI I've noticed changes in cognition within just a couple of days, which precede the mood changes. One is in depressive nihilism, the tendency to think that nothing matters because we're all going to cease existing, which I get (minimally) every winter. This is interesting to me in light of the global reduction in significance some people report. The SSRI I take enables me to care about things *besides* the fact that everybody is going to die, perhaps by making that particular fact less significant to me. Maybe what manifests as increased indifference for some is necessary for others to disentangle knots in their landscape of mattering that are distorting their sensitivity to everything else, by giving a little slack. Some people might need that slack to keep more knots from accumulating, others might only need to shake out one.
The other domain is social cognition. I'm intrigued by the idea that SSRIs might directly or indirectly impact something like an internal status thermostat. A reduction in negativity bias could accomplish the same thing, though (and while that's a super general mechanism, at least it's an actual mechanism--I always felt neurotransmitter hypotheses were pseudo-explanatory, insofar as the relationship between the neurotransmitter and the pathological state is just kinda stipulated).
I think many bipolar people (and also some with MDD) have what you could call chronic inter-episode dysthymia--persistent issues with motivation, energy, and the other variables that rise and fall with the dramatic mood episodes--that doesn't always improve when mood episodes are under control. SSRIs just don't seem to help much with this, and, again anecdotally, SNRIs and bupropion seem to work better for acute-unsticking-from-a-stuck-place than for chronic-resetting-to-a-more-functional-default-place.
I wonder how the two contender hypotheses can be reconciled with the following findings:
Re shifts in emotional processing:
The line of research on acute or sub-acute effects of SSRIs on facial expression recognition or emotional processing produced quite inconsistent findings, complicating any interpretation as possible mediator. A prediction algorithm for antidepressant response using these markers of emotional processing barely was able to predict antidepressant response better than chance, even while taking improvement in symptoms within the first week into account (see https://www.nature.com/articles/s41386-021-00981-z and its supplement).
Re cognitive flexibility and neural plasticity:
Since stressful living conditions are associated with higher levels of depression how would the creation of a "window of plasticity" be desirable in the first place? Would it not simply amplify the toll of the maladaptive environment? Also, I would assume that introducing a "window of change" would widen the range of outcomes under antidepressant treatment compared to placebo, however I am not aware of any statistical manifestations of this increased variance. Rather, the variance between antidepressant and placebo arms in RCTs seems to be identical (see e.g. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0241497)
The study by Tang et al. I find very interesting, I did however not find much other literature replicating or refuting the results. I wonder though how an increase in neuroticism and the accompanying buffering against stressful conditions aligns with a proposed "window of change"
Thank you Benjamin! These are great comments.
The poor performance of markers of emotional processing doesn't reflect well, I agree. I don't know if that is enough for us to say that this hypothesis is likely false. Research on effects of serotonin (e.g. recent study which investigated direct release of serotonin) does seem to suggest a role for emotional processing.
Cognitively flexibility... it's still not clear whether this should necessarily translate into a wider range of outcomes under antidepressant treatment. Here's what Page et al. 2024 say about it:
"The neuroplastic benefits of antidepressants are likely enhanced when they occur in the context of positive experiences in the patient’s everyday environment. People living in more favorable conditions (e.g., employed, college educated, high income) may be more likely to benefit from an antidepressant alone than those not living in favorable conditions [91]. Similarly, psychotherapy, which provides a supportive environment to facilitate positive changes in thinking, feeling, and behaving, enhances the therapeutic potential of antidepressants [92]. In this way, antidepressants can be thought of as opening a window of plasticity that allows the benefits of a positive environment to take hold in neural circuits. By contrast, if experiences in the environment are negative, rodent studies suggest that antidepressant treatment may even be detrimental because enhanced neuroplasticity in an adverse situation facilitates learning that the environment is harmful [93, 94]. However, not all findings support the notion that antidepressants simply make people more susceptible to their environments: rodent studies also show that antidepressant administration prior to stress exposure can buffer against the effects of adversity [95, 96]. Therefore, antidepressants may confer resilience in the face of negative experiences in addition to their ability to enhance the benefits of a positive environment." (https://www.nature.com/articles/s41380-024-02625-2 )
The variance literature is interesting. I don't understand the methodology well enough to know whether the variance issue is considered settled. Some researchers I asked a while back say they weren't convinced yet that variance is identical. And the Stone et al. 2022 (https://www.bmj.com/content/378/bmj-2021-067606 ) showed a greater likelihood of large response in antidepressants vs placebo, which sounds as if it should manifest as increased variance, so I am not sure why it doesn't show up that way.
Yeah, as far as I know, no one has attempted a replication of the neuroticism trial. Would be great if they did!
What are your thoughts on how antidepressant work?
I really appreciate this post, both in terms of your analysis and synthesis and the sources and links. Following up on your last paragraph, any thoughts on what to do with the term "anti-depressants"? Is there a term that does justice to the possible treatment mechanisms or symptoms/traits treated? (a term for communicating with patients and the lay public). Or given the complexity, is it better to avoid such a term? Also, is there a one-sentence explanation of how serotonergic drugs work that you provide to your patients? Or do you describe several possible accounts?
Thanks Thomas! Since antidepressant is the more familiar term to many in the public, I often use it. I try to convey two things: one, they don’t fix any “chemical imbalance,” and are just helping manage symptoms better; two, they are thought to create positive shifts in how brain processes emotions but the actual mechanisms are more complicated and still being worked out. For folks interested in more, I go into further details.
As a non-expert on this stuff, I tend to tell people this when it comes up:
- Yes there's research that shows that various psychiatric medication affects various neurotransmitters and that this can help people.
- But that research is always about a hundred times more complicated than the popular picture you've read in memes and on popular websites.
Really interesting update/summary, thank you. I'm intrigued by the paroxetine-neuroticism paper and would be interested in what a statistician (or other people) would think about the "controlling" for neuroticism. I looked up the NEO questions and they are things like "I am seldom sad or depressed" and "I rarely feel lonely or blue." When the two sets of questions are so similar, how can we interpret this? The paper doesn't investigate individual item scores on the HAM-D or NEO (sample size too small perhaps) but that might be interesting.