Tag: pattern recognition

IQ and mental health

It’s possible that I’ve written about this before, but I’m too lazy to check. I just saw this tweet by Baal about what he calls the “Aaron Swartz syndrome” (of brilliant people lost to mental illness because they put too much pressure on themselves).

(and yay, tweets are publicly visible again)

Baal’s tweet here is about a mutual classmate who we lost over a decade ago.  And this tweet triggered off a thought that I’ve had regarding pattern recognition, and which I might have written about earlier.

Fundamentally, what makes us intelligent is our ability to see patterns. Before the advent of modern “advanced linear algebra”, the difference between giving instructions to a human and to a computer was that the latter had to be incredibly specific. The human, on the other hand, could get approximate instructions, and then quickly see patterns in what they were observing, and get the job done.

Even a lot of “advanced linear algebra” works the same way. You give it a bunch of data, and it uses some mathematical transformations to “learn patterns” about the data, and then looks for these patterns in hitherto unseen data to make predictions. So what makes “artificial intelligence” intelligent is that it can use maths to divine patterns.

I remember taking this Mensa test when in college. It was all about pattern recognition. Four images given, and you need to figure the best fifth image to complete the sequence. That sorts. And Mensa claims to be a “club for the insanely intelligent”, and they use pattern recognition as a means to identify the more intelligent humans.

I can go on but I think I’ve provided sufficient evidence arguments on how intelligence is basically about pattern recognition. The more intelligent you are the better you are at identifying patterns.

Now what does it have to do with mental health?

The answer lies in false positives.

The problem with being good at pattern recognition is that sometimes you can tend to overdo it. You start seeing patterns that don’t really exist. I must mention here that I got over my extremely long-term and fairly deep depression back in 2012 when I was asked to deliver a few lectures on logical reasoning – explaining to my lecturees that correlation does not imply causation convinced me of the same, and I started feelingbetter.

So – because you are good at pattern recognition, you end up seeing too many patterns. I remember this from business school – I saw a bunch of people eating lunch together and thought I’ll go eat with them. And then I noticed a pattern among the set of people (something silly to the effect of “they are all from Section A, and taking this marketing elective”) that didn’t apply to me. And suddenly I decided I didn’t belong there and didn’t go to sit with them.

On that day I remember this happening multiple times, and I finally ate my lunch alone. Now thinking back, this was silly of me – and I had voluntarily brought upon myself unpleasant thoughts (“I don’t belong in this group”) and loneliness.

This is just one example – such things regularly happened through the decade of the 2000s. I would see demons (patterns, basically) where none existed. I would overthink decisions like crazy. I would bring loneliness upon myself. I would make random correlations, that would only serve to depress (“oh, my lucky shirt hasn’t dried, so I won’t be able to do well in today’s exam” types).

Generalising – what you see is that the better you are at seeing patterns, the more the spurious patterns you see (in advanced linear algebra, we call this “overfitting”). And these spurious patterns end up affecting you, and clouding your judgment. And making you less capable of leading life.

I keep thinking, and saying, that my engineering class has been especially badly affected by mental illness. In the class of 30 odd, we’ve lost two people to suicide already (including the person Baal mentioned in his tweet), and know of several others who had mental illness severe enough to either drop out, or take semesters off.

And given that the class was largely made up people from the extreme right tail of the distribution in a highly competitive entrance exam, I’m coming to believe that correlation exists – all of us being superior pattern recognisers, have been prone to recognising spurious patterns, and many have fallen prey to mental illness, to different extents.

PS: I found one blogpost I’ve written about this topic

 

MENSA

Vinod Ganesh is popularly known as MENSA, in Chennai quizzing and other circles. He attained his MENSA membership sometime in 2003-04. The exam (yeah, since it’s a high IQ society, you need to pass an exam to join) was sometime in late 2003 or early 2004, and the results arrived during Saarang 2004. Thinking back, there is a possibility that the nickname could have been mine (though “Wimpy” was well-established by then). I’d also taken the same exam on the same day as Vinod did, and had cracked it. It remains one of the turning points in my life.

I was studying Computer Science at IIT Madras, and was in my final year of the course. Most of the class wanted to go to the US to do their masters, and along came a rumour (possibly substantiated given how universities in America work) that membership of elite clubs such as MENSA was a good bullet point that might enable admission, and offers of aid. Most of my classmates had signed up enthusiastically. The rumour had misled me, in the sense that I had assumed there was little to the exam apart from a bullet point for foreign apps, and had stayed away.

It was a Saturday, and the entrance test was going to happen over three sessions. MENSA entrance is one of those tests where they “recycle” question papers – the papers are taken back at the end of the test, and given out to the next batch. The nature of questions allows them to do this – they are mostly pattern recognition, and are quite hard to “describe” in the absence of the question paper. Sometimes someone else who took the test prior to you would have made marks on the question paper, but it is best you disregard them, for you never know how well they’ve done.

Friends who had written the test in the first batch told me that it was a tough exam. That it was all about pattern recognition and stuff. They also mentioned that for the third session, seats weren’t filled up and they were still taking on-the-spot registrations. I think the entrance fee was a hundred bucks or so, and I made a spur of the moment decision to write the test.

IIT was a hard time for me. For most of my time there, my confidence was at an all-time low. Except for one term, I never did well in academics. Extra curricular activities also floundered, and I would find myself wasting phenomenal amounts of time. I had developed a fear that I wasn’t good enough, and it was feeding onto itself and making things worse. Given my indifferent performances both in class and outside, my peers, too, didn’t have too much respect for me (IIT is strictly meritocratic that way, I must tell you), and that only contributed to my self-doubt. Given that I was going to graduate soon, I knew I needed a stimulus to break out of my rut, and so far hadn’t figured a way out.

MENSA, the exam that I had enrolled for in the last minute, unexpectedly proved to provide the stimulus. It turned out that in my entire Computer Science class (most of whom were double digit rankers in the IIT-JEE, and half of whom had better CGPAs than me), I was the only person to have qualified the MENSA test. I remember a couple of others coming close. Most, including a number of the top rankers in class, hadn’t even come close to qualifying. If my confidence levels were higher earlier, I might have yelled out a “howzzat”. In the event, I didn’t require it, since the success in the exam was enough of a stimulus for me to do well in CAT, which followed, and generally break out of the rut.

In the event, I ended up not joining MENSA. I got a letter asking me to come for a welcome party, where I had to pay a fee to become a lifetime member of MENSA Chennai. I knew I was going to move out of Chennai in about three months’ time, and I thought it would be a waste to become a life member of the Chennai chapter. I remember writing to the Bangalore chapter after I moved back, but the responses were vague, and I never joined. That letter from MENSA which declares my success in the examination, though, sits proudly in my “certificates folder”. And for some three years hence, the fact that I had cracked the MENSA entrance test had adorned my resume.

I’ve never been an “RG” (IIT term for someone who doesn’t hesitate to pull others back in order to get ahead of them), but in this one situation, I had taken great pleasure in my classmates’ failure to qualify for MENSA. For a good reason, I think, since that was responsible in setting me off on a successful run that would last close to two years.

The Trouble With Analyst Reports

The only time I watch CNBC is in the morning when I’m at the gym. For reasons not known to me, my floor in office lacks televisions (every other floor has them) and the last thing I want to do when I’m home is to watch TV, that too a business channel, hence the reservation for the gym. I don’t recollect what programme I was watching but there were some important looking people (they were in suits) talking and on the screen “Target 1200” flashed (TVs in my gym are muted).

Based on some past pattern recognition, I realized that the guy in the suit was peddling the said stock (he was a research analyst) and asking people to buy it. According to him, the stock price would reach 1200 (I have no clue what company this is and how much it trades for now). However, there were two important pieces of information he didn’t give me, because of which I’ll probably never take advice from him or someone else of his ilk.

Firstly, he doesn’t tell me when the stock price will reach 1200. For example, if it is 1150 today, and it is expected to reach 1200 in 12 years, I’d probably be better off putting my money in the bank, and watching it grow risk-free. Even if the current price were lower, I would want a date by which the stock is supposed to reach the target price. Good finance implies tenure matching, so I should invest accordingly. If the stock is expected to give good returns in a year, then I should put only that money into it which I would want to invest for around that much time. And so forth.

Then he doesn’t tell me how long it will stay at 1200. I’m not an active investor. I might check prices of stocks that I own maybe once in a week (I currently don’t own any stock). So it’s of no use to me if the price hits 1200 some time during some intraday trade. i would want the price to remain at 1200 or higher for a longer period so that I can get out.

Thirdly and most importantly, he doesn’t tell me anything about volatility. He doesn’t give me any statistics. He doesn’t tell me if 1200 is the expected value of the stock, or the median, or the maximum, or minimum, at whatever point of time (we’ve discussed this time bit before). He doesn’t tell me what are the chances that I’ll get that 1200 that he professes. He doesn’t tell me what I can expect out of the stock if things don’t go well. And as a quant, I refuse to touch anything that doesn’t come attached with a distribution.

Life in general becomes so much better when you realize and recognize volatility (maybe I’ll save that for another discourse). It helps you set your expectations accordingly; it helps you plan for situations you may not have thought of; most importantly it allows you to recognize the value of options (not talking about financial options here; talking of everyday life situations). And so forth.

So that is yet another reason I don’t generally watch business TV. I have absolutely no use for their stock prediction and tips. And I think you too need to take these tips and predictions with a bit of salt. And not spend a fortune buying expensive reports. Just use your head. Use common sense. Recognize volatility. And risk. And you’ll do well.

Remembering Names and Pattern Recognition

I spent the first half of this week attending a Pan-Asia training program in Hong Kong. Most of the people attending this program were from the Tokyo and Hong Kong offices of our firm, and most of them happened to be natives of China, Japan and Korea. It was a wonderful training program and gave much scope for networking. The biggest surprise to me, however, was about how bad I was during the two days at remembering names – something I consider myself good at.

We Indians constantly crib that westerners are usually bad at catching our names while on the other hand that we don’t have much trouble remembering their names. Thinking about it, I think name recognition is basically an exercise in pattern recognition and the ease of rememberance of a certain class of names depends on how easily we can recognize those patterns.

If you are familiar with the broad class of names of a particular ethnicity (let’s say Indian Hindu for example), you don’t really need to remember the name as a collection of syllables. You only need to know say the first letter, or an abstract concept which is what the name means, or a combination of this and it is likely that you can remember the full name.

The thing wiht western names, however, is that due to Hollywood, or sport, or colonial rule, or the fact that Indian Christians have names similar to mostly christian Westerners, most Western names are familiar to us. And because of this familiarity, it is not hard at all for us to remember the name of the average Westerner. On the contrary, due to lack of exposure, Westerners can’t recognize patterns in Indian names because of which it is hard for them to remember our names.

It is due to lack of general familiarty with Chinese and Japanese names that I found it so hard to remember names during the recent trip. There was no way I could break down names into easy combination of syllables (yeah for example Hi-ro-hi-to consists of all easy syllables, but how many people called hirohito would you know for you to remember the whole name by just remembering part of the name) and so I had the additional responsibility of remembering all the syllables in the names and the combination in which they occurred.

On a related note, a disproportionate proportion of people of Chinese origin at the training had a christian (western) first name and a chinese last name (eg. Michael Chang). But then I suppose this is because a lot of Chinese people adopt a “Western name” to make matters simple when they migrate or something (so for example, someone called Chang Sun-Wang will convert his name to Stephen Chang).