The promise of mind reading techniques whether it is a former FBI analyst or one of Paul Ekman’s microexpression reading human lie detectors. I become aware of this cottage industry during every trial-by-media where suspicion piles upon someone not yet charged with murder.
I have to admit I am skeptical that anyone has such an amazing power to see through the facade of a stranger and with a greater-than-chance determine if they are telling the truth or not. Doubly so because I am someone who is constantly misinterpreted, I have to manage my gestures and facial expressions because my confusion is often misread as disagreement; my approval for disapproval; even a simple statement like “I’m not hungry right now” is wrongly generalized as not liking the particular cuisine… and not that I just don’t want to eat anything right at this moment.
However if placed under the microscope by one of these former FBI body language experts would I feel a intense sense of validation ? Would I exclaim “yes, I feel seen, heard… you get me!”?
I have no doubt some people are more perceptive about emotional nuances than others: film and theatre actors who are trained to observe and mimic, people who have grown up in abusive or emotionally unstable households and are hyper sensitive to small changes in the mood of others (which of course may make them prone to more ‘false positives’ and paranoia), and of course mentalists like cold readers and palmists.
However being more emotionally perceptive doesn’t necessarily mean you can tell if someone is lying—or a particular statement is false, especially if that person is especially good at telling the truth, or like me—their natural body language and expression doesn’t express what you’d expect.
What I have greater faith in is that given even a small but emblematic example of a person’s extemporaneous speech you could derive an accurate personality and world-view portrait of them. In the same way that an accent can help you pinpoint the geographical and economic origin of a person (think of comedies like The Nanny that play up on this convention). Harry Shearer once explained that to play Richard Nixon he channeled Jack Benny—believing that Nixon’s persona and particularly his way of telling jokes was consciously or unconsciously modelled on that of Benny. Likewise Vladimir Putin’s distinctive gait has been attributed to a prenatal stroke, or that his subordinates including Dmitry Medvedev have “copied the boss”, the more persuasive explanation is that they all picked up the habit from watching Soviet Spy films as youngsters and wanting to emulate the hero.
The kinds of films, television, and role models, books, music and lyrics that someone has absorbed would also influence or at least be indicative of their world view. Given enough of these tells, while I am not sure that you could tell if someone is or isn’t a murderer, you could certainly gain a accurate insight into their worldview, the mental models they have about the world, what they value, what their ethics system is like etc. etc.
How much information can you extract about a person from a written transcript that they aren’t aware they are sharing is probably startling, but rarely or predictably “he’s a murderer” level.
On the scale between “pseudoscience that provides either completely random results or exactly what its operator wants to hear” and “always provides the correct answer”, there are some uncomfortable points where we probably get first, such as “provides the correct answer 99% of the time” (and with the 1% chance you are unlucky, and you are screwed because no one is going to believe you) or “provides the correct answer for neurotypical people” (and if you are an autist, you are screwed).
I am someone who is constantly misinterpreted… even a simple statement like “I’m not hungry right now” is wrongly generalized as not liking the particular cuisine
Similar here. My face expression is always on “neutral”, and my statements, no matter how simple and literal, are often creatively interpreted. And I guess I am sufficiently unusual, so heuristics like “let’s assume that he thinks/feels what an average person would think/feel in this situation” also fail. It took me a lot of time to understand myself to the level where I can explain things about myself verbally, but when I do, people usually find it implausible and try to find some hidden meaning behind my words.
So… a machine that could read my thoughts could feel validating. Assuming it does so correctly. But there is also a chance it would provide correct answers for most people, and incorrect answers for the few unusual ones.
On the scale between “pseudoscience that provides either completely random results or exactly what its operator wants to hear” and “always provides the correct answer”, there are some uncomfortable points where we probably get first, such as “provides the correct answer 99% of the time” (and with the 1% chance you are unlucky, and you are screwed because no one is going to believe you) or “provides the correct answer for neurotypical people” (and if you are an autist, you are screwed).
I’m afraid I need you to rephrase or elaborate on what you meant by this—are you saying, aware of a technique or method which is right 99% of the time or thereabouts. Or are you saying human variability makes such a technique impossible for anything but the most narrow populations? Or have I likely (and in a meta-way appropriately) completely missed the point? What do you think of more generally—as I explicate in the second half—revelations about a person’s internalized belief structures, including their hero’s and related moral system, but also the idea of idiolect being a symptom of their thinking and model of the world even if it is not a mechanism for directly ascertaining their personal belief in this or that specific statement?
Oops, I actually misinterpreted one part—when you wrote “if placed under the microscope”, my brain interpreted this literally, as if you were talking about a hypothetical future version of “mind reading” that would include checking your neurons by a microscope and probably interpreting the results using an AI.
What I meant is that people usually think about these things in “yes or no” categories. For example, if you asked people whether existing lie detectors work, the most frequent answers would probably be “of course yes, why would they use them otherwise?” or “of course not, it’s total bunk”.
There I didn’t mean to make a statement about lie detectors per se, but about: this is how people think about technologies when you ask them. They think the correct answer is either “yes” or “no”, even if it is something complicated like “sometimes” or “yes, but with exceptions”. If the popular belief happens to be an unqualified “yes”, and you happen to be the exception, you are screwed.
I believe the current “mind-reading” techniques like Paul Ekman’s are hit and miss. That they probably often work with typical people in typical situations, but fail when something unusual happens. (Someone may be scratching their nose because they are lying, but sometimes the nose is just itchy for a completely unrelated reason. Or the person is lying, but in a different way than you assume. Or is just generally uncomfortable, maybe thinking “this is true, but they seem unlikely to believe me”.)
Practically, “films, television, and role models, books, music and lyrics that someone has absorbed” are an enormous amount of data, especially for people who consume a lot of these media. Maybe someone who reads one book in ten years and only watches the mainstream TV could be modeled this way. But if you asked me to give you a list of books I have read and the movies I have seen, I could probably remember only a small fraction of them. How specifically is Paul Ekman going to find out which movies I have downloaded from internet, watched in private, then deleted and forgot about them? Is he really going to read the LW Sequences, and Worm, and other extremely long texts written for tiny subcultures, just to get a better model of me? No way. -- However, with the help of an AI processing the tons of texts could become feasible. (The problem of figuring out who read what still remains.)
You probably could figure out some things, for example you could notice that people are, maybe unknowingly, repeating some sentences or argument structures from some book or movie heroes. Then it would make sense to assume that they somehow identify with the heroes and the values they represent. But even then, people sometimes adopt some values differently than they were originally meant. For example, someone could be a 99% Randian, but believe that Rand made a mistake about one specific topic. But maybe the topic is relevant to the thing the mind-reader is trying to figure out. Also it’s possible that someone was a Randian (sorry for repeating the same example) in the past, but is not anymore, but still keeps some verbal manners or something. In other words, you might figure out the sources that influenced the person… but not whether the person has adopted that perspective wholesale or just partially, and how their opinions developed over time.
This will probably work better for some kinds of people, and worse for others. For example, many people, if they join a movement, they adopt 100% of the movement’s beliefs, because they really strongly want to belong. But other people are just like “yeah, they make some really good points, but are mistaken about many things”. Modelling the latter would be much more difficult. I assume that being difficult to model will positively correlate with intelligence and autism. (As in: intelligent people have more complex and therefore less predictable models of the world; autists care less about being compatible with their group and therefore knowing which groups they belong to has smaller predictive power.)
I am overwhelmingly confident that analysis of the kinds of narratives that a particular person spins, including what tropes they evoke—even if you’re not familiar with the tropes previously—would reveal a lot about their worldview, their ethical structure, the assumptions and modelling they have about how people, institutions, and general patterns they believe underlay the world.
A oversimplified example is a person who clearly has a “victim “mentality” and an obsession with the idea of attractiveness because they always use sentence structures (i.e. “they stopped me”) and narratives where other people have inhibited, bullied, envied, or actively sought to stifle the person telling the story and these details disproportionately make reference to people’s faces, figures, and use words like “ugly” “hot” “skinny” etc. It is not necessary to know what films, books, periodicals they read.
I think you would get the set of topics, but not necessarily the right idea about how exactly those topics apply to the current situation. To use your example, if someone’s speech patterns revolve around the topic of “bullying”, it might mean that the person was bullied 50 years ago and still didn’t get over it, or that the person is bullied right now, or perhaps that someone they care about is bullied and they feel unable to help them. (Or could be some combination of that; for example seeing the person they care about bullied triggered some memories of their own experience.)
Or if someone says things like “people are scammers”, it could mean that the person is a scammer and therefore assumes that many other people are the same, or it could mean that the person was scammed recently and now experiences a crisis of trust.
This reminds me of an anime Psycho Pass, where a computer system detects how much people are mentally deranged...
...and sometimes fails to distinguish between perpetrators and their victims, who also “exhibit unusual mental patterns” during the crime; basically committing the fundamental attribution error.
Anyway, this sounds like something that could be resolved empirically, by creating profiles of a few volunteers and then checking their correctness.
To use your example, if someone’s speech patterns revolve around the topic of “bullying”, it might mean that the person was bullied 50 years ago and still didn’t get over it
Yes. Which is invaluable information about how they see the world currently. How is that not the ‘right idea’? If that is how they continue to currently mentally represent events?
Your ‘people are scammers’ example is irrelevant, what is important is if they constantly bring in tropes or examples or imply deception. They may never use the word ‘scammer’ ‘mistrustful’ or make a declaration like ‘no one has integrity’. The pattern is what I’m talking about.
The promise of mind reading techniques whether it is a former FBI analyst or one of Paul Ekman’s microexpression reading human lie detectors. I become aware of this cottage industry during every trial-by-media where suspicion piles upon someone not yet charged with murder.
I have to admit I am skeptical that anyone has such an amazing power to see through the facade of a stranger and with a greater-than-chance determine if they are telling the truth or not. Doubly so because I am someone who is constantly misinterpreted, I have to manage my gestures and facial expressions because my confusion is often misread as disagreement; my approval for disapproval; even a simple statement like “I’m not hungry right now” is wrongly generalized as not liking the particular cuisine… and not that I just don’t want to eat anything right at this moment.
However if placed under the microscope by one of these former FBI body language experts would I feel a intense sense of validation ? Would I exclaim “yes, I feel seen, heard… you get me!”?
I have no doubt some people are more perceptive about emotional nuances than others: film and theatre actors who are trained to observe and mimic, people who have grown up in abusive or emotionally unstable households and are hyper sensitive to small changes in the mood of others (which of course may make them prone to more ‘false positives’ and paranoia), and of course mentalists like cold readers and palmists.
However being more emotionally perceptive doesn’t necessarily mean you can tell if someone is lying—or a particular statement is false, especially if that person is especially good at telling the truth, or like me—their natural body language and expression doesn’t express what you’d expect.
What I have greater faith in is that given even a small but emblematic example of a person’s extemporaneous speech you could derive an accurate personality and world-view portrait of them. In the same way that an accent can help you pinpoint the geographical and economic origin of a person (think of comedies like The Nanny that play up on this convention). Harry Shearer once explained that to play Richard Nixon he channeled Jack Benny—believing that Nixon’s persona and particularly his way of telling jokes was consciously or unconsciously modelled on that of Benny. Likewise Vladimir Putin’s distinctive gait has been attributed to a prenatal stroke, or that his subordinates including Dmitry Medvedev have “copied the boss”, the more persuasive explanation is that they all picked up the habit from watching Soviet Spy films as youngsters and wanting to emulate the hero.
The kinds of films, television, and role models, books, music and lyrics that someone has absorbed would also influence or at least be indicative of their world view. Given enough of these tells, while I am not sure that you could tell if someone is or isn’t a murderer, you could certainly gain a accurate insight into their worldview, the mental models they have about the world, what they value, what their ethics system is like etc. etc.
How much information can you extract about a person from a written transcript that they aren’t aware they are sharing is probably startling, but rarely or predictably “he’s a murderer” level.
On the scale between “pseudoscience that provides either completely random results or exactly what its operator wants to hear” and “always provides the correct answer”, there are some uncomfortable points where we probably get first, such as “provides the correct answer 99% of the time” (and with the 1% chance you are unlucky, and you are screwed because no one is going to believe you) or “provides the correct answer for neurotypical people” (and if you are an autist, you are screwed).
Similar here. My face expression is always on “neutral”, and my statements, no matter how simple and literal, are often creatively interpreted. And I guess I am sufficiently unusual, so heuristics like “let’s assume that he thinks/feels what an average person would think/feel in this situation” also fail. It took me a lot of time to understand myself to the level where I can explain things about myself verbally, but when I do, people usually find it implausible and try to find some hidden meaning behind my words.
So… a machine that could read my thoughts could feel validating. Assuming it does so correctly. But there is also a chance it would provide correct answers for most people, and incorrect answers for the few unusual ones.
I’m afraid I need you to rephrase or elaborate on what you meant by this—are you saying, aware of a technique or method which is right 99% of the time or thereabouts. Or are you saying human variability makes such a technique impossible for anything but the most narrow populations? Or have I likely (and in a meta-way appropriately) completely missed the point? What do you think of more generally—as I explicate in the second half—revelations about a person’s internalized belief structures, including their hero’s and related moral system, but also the idea of idiolect being a symptom of their thinking and model of the world even if it is not a mechanism for directly ascertaining their personal belief in this or that specific statement?
Oops, I actually misinterpreted one part—when you wrote “if placed under the microscope”, my brain interpreted this literally, as if you were talking about a hypothetical future version of “mind reading” that would include checking your neurons by a microscope and probably interpreting the results using an AI.
What I meant is that people usually think about these things in “yes or no” categories. For example, if you asked people whether existing lie detectors work, the most frequent answers would probably be “of course yes, why would they use them otherwise?” or “of course not, it’s total bunk”.
There I didn’t mean to make a statement about lie detectors per se, but about: this is how people think about technologies when you ask them. They think the correct answer is either “yes” or “no”, even if it is something complicated like “sometimes” or “yes, but with exceptions”. If the popular belief happens to be an unqualified “yes”, and you happen to be the exception, you are screwed.
I believe the current “mind-reading” techniques like Paul Ekman’s are hit and miss. That they probably often work with typical people in typical situations, but fail when something unusual happens. (Someone may be scratching their nose because they are lying, but sometimes the nose is just itchy for a completely unrelated reason. Or the person is lying, but in a different way than you assume. Or is just generally uncomfortable, maybe thinking “this is true, but they seem unlikely to believe me”.)
Practically, “films, television, and role models, books, music and lyrics that someone has absorbed” are an enormous amount of data, especially for people who consume a lot of these media. Maybe someone who reads one book in ten years and only watches the mainstream TV could be modeled this way. But if you asked me to give you a list of books I have read and the movies I have seen, I could probably remember only a small fraction of them. How specifically is Paul Ekman going to find out which movies I have downloaded from internet, watched in private, then deleted and forgot about them? Is he really going to read the LW Sequences, and Worm, and other extremely long texts written for tiny subcultures, just to get a better model of me? No way. -- However, with the help of an AI processing the tons of texts could become feasible. (The problem of figuring out who read what still remains.)
You probably could figure out some things, for example you could notice that people are, maybe unknowingly, repeating some sentences or argument structures from some book or movie heroes. Then it would make sense to assume that they somehow identify with the heroes and the values they represent. But even then, people sometimes adopt some values differently than they were originally meant. For example, someone could be a 99% Randian, but believe that Rand made a mistake about one specific topic. But maybe the topic is relevant to the thing the mind-reader is trying to figure out. Also it’s possible that someone was a Randian (sorry for repeating the same example) in the past, but is not anymore, but still keeps some verbal manners or something. In other words, you might figure out the sources that influenced the person… but not whether the person has adopted that perspective wholesale or just partially, and how their opinions developed over time.
This will probably work better for some kinds of people, and worse for others. For example, many people, if they join a movement, they adopt 100% of the movement’s beliefs, because they really strongly want to belong. But other people are just like “yeah, they make some really good points, but are mistaken about many things”. Modelling the latter would be much more difficult. I assume that being difficult to model will positively correlate with intelligence and autism. (As in: intelligent people have more complex and therefore less predictable models of the world; autists care less about being compatible with their group and therefore knowing which groups they belong to has smaller predictive power.)
I am overwhelmingly confident that analysis of the kinds of narratives that a particular person spins, including what tropes they evoke—even if you’re not familiar with the tropes previously—would reveal a lot about their worldview, their ethical structure, the assumptions and modelling they have about how people, institutions, and general patterns they believe underlay the world.
A oversimplified example is a person who clearly has a “victim “mentality” and an obsession with the idea of attractiveness because they always use sentence structures (i.e. “they stopped me”) and narratives where other people have inhibited, bullied, envied, or actively sought to stifle the person telling the story and these details disproportionately make reference to people’s faces, figures, and use words like “ugly” “hot” “skinny” etc. It is not necessary to know what films, books, periodicals they read.
I think you would get the set of topics, but not necessarily the right idea about how exactly those topics apply to the current situation. To use your example, if someone’s speech patterns revolve around the topic of “bullying”, it might mean that the person was bullied 50 years ago and still didn’t get over it, or that the person is bullied right now, or perhaps that someone they care about is bullied and they feel unable to help them. (Or could be some combination of that; for example seeing the person they care about bullied triggered some memories of their own experience.)
Or if someone says things like “people are scammers”, it could mean that the person is a scammer and therefore assumes that many other people are the same, or it could mean that the person was scammed recently and now experiences a crisis of trust.
This reminds me of an anime Psycho Pass, where a computer system detects how much people are mentally deranged...
...and sometimes fails to distinguish between perpetrators and their victims, who also “exhibit unusual mental patterns” during the crime; basically committing the fundamental attribution error.
Anyway, this sounds like something that could be resolved empirically, by creating profiles of a few volunteers and then checking their correctness.
Yes. Which is invaluable information about how they see the world currently. How is that not the ‘right idea’? If that is how they continue to currently mentally represent events?
Your ‘people are scammers’ example is irrelevant, what is important is if they constantly bring in tropes or examples or imply deception. They may never use the word ‘scammer’ ‘mistrustful’ or make a declaration like ‘no one has integrity’. The pattern is what I’m talking about.