It’s 3 am. You’re exhausted but continue to type away on some thread. Your significant other wakes up and attempts to get you to come to bed, but you can’t. Someone is wrong on the internet, and it’s your job to set them straight.
This was me in the past as well. I’m sure many of you have gone through this phase to the other side, eventually coming to the conclusion that you just don’t argue with stupid (or some other flavor of this conclusion). Maybe now you see some ridiculous comment, and you just suppress the initial irritation and scroll on. That’s fine. I’m not here to convince you that arguing with strangers on the internet is a productive use of your time.
What I would like to do, is explain the overall architecture of the brain. Why is that relevant? Well, once you understand the brain's basic architecture, a few things become crystal clear.
Purposely changing someone’s mind
- is very rarely impossible.
- almost never happens in a single conversation.
- requires an understanding of their current perspective
- often requires changing their physical location
Once you are familiar with the brain’s structure, the reasons behind these points become obvious. You’ll even be able to deduce other key points depending on the person.
Before I dive too deep into the weeds, I'd just like to confirm that I am not an emotionless, manipulative robot, but the way in which I'm going to be talking about different situations in this post will definitely come off that way. Clearly, anyone planning on using the information I'm going to present here should use some tact when talking through actual topics with real people. It should also be mentioned that any heavy-handed use of these techniques is not only likely to be ineffective, but might also have the opposite effective. People don't like to feel manipulated, so either get some buy in, or be subtle.
Super Basic Structural Overview
Let’s start with the concept of a neural network. The ones in your brain are much more complicated than the one I’m about to show you, but the basic concepts are widely applicable to all neural networks.
A neural network can be thought of as a series of weighted paths between nodes. There are input nodes, hidden nodes, and output nodes. In this example, there's only one layer of hidden nodes, and all of the path's go in one direction. Neither of these things are true in the vast majority of useful neural networks, but for the sake of simplicity, we'll just use this example.
Imagine a chunk of data flowing through the diagram. When it's at a node, it has to choose which arrow it will follow. This choice is often random to start with (say from input to hidden each arrow is weighted at 25%), but then the key piece of this structure comes in.
As a path gets more use, its weight increases, which makes it more likely that similar input data will travel down the same paths in the future. At some point, certain inputs will converge on definitive paths. Doing this process with defined inputs and expected outputs is called training. In order to change the weights of paths in a trained neural network, the training data has to change.
What does this mean with respect to a brain? Let’s ask a series of questions that’ll hopefully generate some insight.
Is it impossible to retrain a neural network?
Is it possible to retrain a neural network by giving it a single set of different inputs?
Is it easier to change neural networks that were trained on smaller sets of data?
How effectively can you retrain a neural network by completely changing its training data?
Hopefully you already see where I’m going with this.
So is it impossible to retrain a neural network? Obviously not. Neural networks would be pretty fucking useless if they couldn't be retrained. Then why did I bring this up? Well, if brains are essentially complicated neural networks, then saying a brain can't be changed is just as silly. Brains change. That's why they are useful. Just thought I'd get this out of the way at the start to dispel the myth that people's minds can't be changed (no matter how long they've been a certain way).
Let's dig a little deeper. Is it possible to retrain a neural network by giving it a single set of different inputs? I think the answer to this is yes, but it’s very, very unlikely unless the network is brand new (and even then, it would be terribly trained) or has a very even distribution of weights between at least two cases. What does this tell us? This basically says that your long-ass dissection of someone's argument on a forum is only likely to influence two groups of people - the ones who have never heard the argument before and the fence-sitters. For everyone else, your post is just a small collection of data points - not enough to change their perspective (unless, of course, they have organized their brain to heavily weight the kinds of data points you presented, but even then, this type of person will end up researching the new information extensively themselves in order to retrain their brain).
This all leads us directly into the next point which is that it is clearly easier to retrain neural networks that were trained on less data. Less initial training data means even a few new data points will have a greater effect. This means that the best approach for changing someone's mind is to use the least trained network. Or stated more simply, novel approaches are more effective than ones that a person has a prepared rebuttal for.
Finally, how effectively can you retrain a neural network by completely changing its training data (inputs with expected outputs)? Very effectively. I would go so far as to say that if you can massively change where someone is getting all of their training data from, then they will quickly become a completely different person.
Alright, so how do I do that?
Well first, let's define what I mean by an input. An input is any source of information that impacts the way you think. Examples would include: advice from a trusted friend or family member, a news station, a random website, your boss, etc. These are your inputs. These things constantly reinforce a specific architecture in your brain. It follows that if one of these sources becomes unreliable, or conflicts with another trusted source, then the architecture of your brain will change to compensate.
Alright, so the key to changing someone’s mind, however indirectly, is by changing their trusted sources. This seems obvious enough, but how do you do this?
Well, there are two major ways. One is to become a trusted source yourself, and the other is to restrict access to their conflicting trusted sources so as to disrupt the constant stream of reinforcing inputs.
What do I do as a trusted source?
As someone’s trusted source, you have some influence over how they think. Generally, this isn’t enough to completely change someone’s mind on any given topic (since many topics will be heavily weighted against you). But it may be enough to influence their trust in some of their other inputs that are reinforcing opposing beliefs.
For this part I like to use the board game RISK as an analogy. For this analogy, the board is a person's mind, and the armies are the various points of view the person has on any given topic. Taking over the world would then mean that your point of view won the 'war' in the person's mind.
To me this analogy is useful for two reasons:
- It shows the importance of frequently reinforcing key points
- It highlights terrible strategies for changing someone's mind (AKA going for a country with a ton of armies on it early on)
The optimal strategies in RISK also happen to be extremely effective techniques for changing someone's mind. If a country is heavily reinforced, it's a bad idea to try and take over nearby countries. You should chip away at the weak points and prevent them from getting bonus reinforcements whenever possible while trying to get as many bonuses as you can. A 'bonus' in this example would be a trusted source which would help to reinforce the point of the controlling army.
So what are your goals? Disrupt the constant reinforcement of certain inputs and defend the bonuses you currently have while avoiding anything that has been heavily reinforced until you've conquered the weakly reinforced areas, and built up enough reinforcements.
In other words, attack from new, novel angles (preferably ones that align with the particular individuals currently held beliefs), and play the long game. You'd better train up your creativity and learn their view inside-out if this is your main game plan.
How do I disrupt constant reinforcement from certain inputs?
The main way is to change locations (and limit internet activity). Obviously, this isn't always feasible. Traveling/Relocating is expensive (and fuck you, I've got to be able to use the internet), but hear me out. Most brains have a large subset of beliefs that are constantly reinforced by the people around them. In most cases, these beliefs seem so normal, that no one thinks twice about them. What would happen then, if the person who held these kinds of beliefs were to move to an area, where their beliefs were extremely unusual and/or ridiculed?
Well in most cases, this would be pretty painful for the person. Retraining a brain consumes a lot of energy, so it's actually painful to start the process (because the brain doesn't want to waste resources by going through constant retraining). So more often than not, a person will default back to their trusted sources despite being farther removed from many of them. This is where restricting internet use comes in.
You might think that the internet would make it incredibly easy for anyone to learn the truth, and in a way, it does, but I think a more accurate description would be that the internet makes it incredibly easy for anyone to learn what they want to learn. If they are cautious and aware of their brain's shortcomings, they are more likely to obtain an accurate view of the world when using the internet. If not, well, people convince themselves of all kinds of weird shit. This is bad if your goal is to change a person's mind, because you'll be fighting against all the shit being shoveled into their brain by random people.
This technique (limiting reinforcing inputs) is easy to be heavy-handed with, but remember that that diminishes it's effectiveness. Give-and-take and group participation go a long way here.
Combine both techniques (restricting access to reinforcing inputs and becoming a trusted input yourself) for the most effective approach.
Alright, now that I've proven myself to be a manipulative psychopath, let me try and redeem myself.
How do I change my mind?
Ready to feel enlightened?
Here's how you can train your brain to make it easy to retrain. Or in other words, here's how you can adjust your beliefs to be more closely aligned with reality.
- Take your most strongly held beliefs and write them down. Try to find any dependencies they have on each other.
- Ask yourself for each belief, 'what would convince me that this belief is wrong?' These will be your 'deal-breakers.' (If you can't come up with an answer to this question, that should be a red flag. Hopefully future steps will help you come up with a few deal-breakers.)
- Actively research opposing views. (This process can be very painful. Why would anyone want to read a bunch of stuff they know is wrong? Well, as I said before, changing your mind is painful, and this is a part of that process.)
- Try and find points in opposing views for which you have no good answers.
- Revisit number two. Are any of the unanswered questions related to your deal-breakers?
- Iterate. Remind yourself that this is a constant process that needs to be done multiple times for it to stick.
Most of the time, this process gets pretty messy and only succeeds in making you more unsure about everything. Good. That's the point. Certainty is a characteristic of an inflexible mind, so if you want to have your beliefs closely resemble reality, you better tailor your degree of certainty to the amount of evidence reinforcing your beliefs (and always have a deal-breaker set in place for even your most cherished ones).
Thank you for reading, and remember to retrain wisely!
As an experiment, I've decided I'll only be upvoting comments with counterpoints. So, if anyone reads something in this post that struck them as terribly misinformed, let me have it.