Mind vs. Machine: Importance of Emotions in predicting Human Behaviour
Humans have come a long way since they first discovered fire. Time has been kind to humans if you think of all the advancements we’ve made since then. From our ancestors engaging in activities to feed their families to now engaging in activities to feed our mind, we have truly been a marvel of evolution.
In keeping with that pace, humans have forever sought to change the way things are done. Take, for example, fire. They discovered that friction causes fire. Fire was useful. Fire provided heat. Fire provided better tasting food. But creating one took too long. I mean what if those tools got wet? After centuries of dealing with this, somewhere in County Durham, in the year 1826, an English Chemist named John Walker decided he’d had enough. And so he invented ‘Matches’. It was the first successful iteration of a friction match simply because he wanted to find an easier method to obtain fire. Just like that, no one ever complained about the wet tools or ground again.
“A machine is anything that reduces human effort. Anything that simplifies work, or saves time, is a machine …” — Ranchoddas Shyamaldas Chanchad
We will always try to find an easier way to get things done faster, better and more efficiently. In the 21st century, there is no doubt as to how we all aim to achieve that. Machines! But since the advent of the Industrial Revolution, the focus of their use has shifted greatly. Where they were first only used for heavy-duty jobs, they are now taking the world by storm by their superb and excellent computational capabilities in every field of study.
Behavioural Sciences and Artificial Intelligence, are burgeoning fields fused together to read, analyse and predict human behaviours with the use of trained machines. Why? Because machines do it better. Or so we thought!
You are only as good as the company you keep!
Machine Learning(ML) engineers spend hours using algorithms to teach a model to learn, analyze and predict behaviours. But the model is only as good as the data that is provided to it. Engineers seek to replicate the workings of a human mind in their models. But it isn’t as simple as feeding ‘yeses’ or ‘nos’ to the machine as ‘0’s and ‘1’s. Any behavioural science expert and even the engineers themselves will tell you that human behaviour cannot be simply divided into a ‘1’ or ‘0’, ‘present’ or ‘absent’, ‘black’ or ‘white’. There is a massive grey area whose existence the machine is oblivious to. Why? you ask again? Because humans haven’t fully understood matters of the grey yet either.
Grey areas matter…
The ‘Grey Matter’ (you see what I did there?), a crucial component of the Brain’s Central Nervous System, form the basis of the structures that function as a medium for interneuronal communication. This is the network of communication through which messages are sent within the brain and to and from different parts of the body. Much like a machine, information that comes in is processed and analysed to produce a logical output. Simple isn’t it? So then where’s the issue?
The steps of information processing in the brain are not really linear or sequential. Consider the information that you are taking in right now as you read. Here’s what is actually happening inside your brain…
Your eyes are looking at black symbols (letters). These are fed in as just that, to the occipital lobe. The adjacent Parietal lobe perceives these to be information and sends it to the Temporal lobe who recognizes these symbols as alphabets and then words from its memory stores. If it comes across a new $ymbol, a quick consultation with the Frontal lobe will help it decide the new symbol’s purpose and identity. (Did you notice the dollar sign above? Or was your interlobular communication that quick?).
Next, your frontal lobe (the executive decision-maker) takes control and quickly analyses the information to form logical conclusions. If smeothing (but you read it as ‘something’ didn’t you?) doesn’t make sense, it quickly retrieves stored data from the temporal lobe to predict an instance. When the information looms in the frontal lobe, data is sent back and forth to assess its credibility and validity.
Additional data might be retrieved from the parietal and occipital lobe to ensure that data is processed correctly, (like when your frontal lobe made you double-check the dollar sign or the word ‘smeothing’ to ascertain its prediction right). That is what is happening right now as you reead. (Made your frontal lobe check again!)
Phew! All that in a few seconds huh? Well, your brain’s not done yet. Like I said. It’s not linear. A major portion of the decision making or problem-solving process in humans involve ‘Emotions’! As much as you’d want to go ‘I’m not like that’ right now, buddy I’m here to tell you, we are all the same. ‘Emotions’ are a byproduct of neurochemical reactions that take place in your brain. Now unless you’re someone whose brain doesn’t produce these, you, my friend are just like the rest of us. Also, the alternative is scarier. Trust me!
Consulting with the Taylor of the brain…Swiftly!
The Cortex, wherein lie the 4 lobes, communicates with the midbrain for matters of what we like to call ‘the heart’. This is where all the mushy-gushy, ooey-gooey, emotions originate. So you can see how a simple consultation with the midbrain can produce an emotional response to any instance. It’s like messing with the bull. You most definitely will get the horns!
All jokes aside, emotions aren’t really as bad as they’re made to be. It’s understandable if the first thing that comes to mind when I speak of emotionally guided decisions, are the ones you’ve taken for your family, partner, friends, etc. But, in reality, every decision we make is emotional.
The ‘emotional’ midbrain acts as a compass that guides us through the decision making process. It is a complex and empathic (understand the plight of others) system made up of experiences, ethics, memories, failures and successes. A decision isn’t simply partnered with an ‘emotion’ and thrown out. It is evaluated using all these factors to arrive at a prediction that ensures the best possible outcome for you, and then and only then is the decision finalized. In fact, emotions help optimize decisions, and its neglect will render your decision meaningless and in most cases unpredictable because you literally haven’t considered it all.
Now, you will be oblivious to this process as it happens, because it’s non-conscious. It is even more difficult to accept this because you may argue that these emotions are never seen on your face. This is the primary reason why teaching a model to read human behaviour is so difficult. A model heavily depends on the change of a ‘capturable’ feature to predict a ‘learnt’ outcome, but when that change isn’t seen, the model assumes that nothing has happened.
Take ‘Smiling’ for example. A model is trained on thousands of data sets to capture and predict the presence of a ‘Smile’ when there’s movement noted around the lips and cheeks. When a person smiles, their lips extend, form a curve, the cheek muscles move upwards, etc, etc, etc. By all reason and logic, this should mean that the person is smiling and is thus happy. But even before I state my case here, you too, know for a fact that just this much is not enough. Was it just a simple curve or did the teeth show? Did the eyes squint or did they open wider? Did the eyebrows move upwards or did they move differently to show different expressions? All this could mean that the person was smiling, but change the combination and you could have a smile that was sarcastic, surprised, amused, confused, nervous, etc, etc, etc.
Now add to that a host of other factors like body language, gestures, tone, speech, head movement, and you’re suddenly looking at a hundred combinations that could mean ten different things just because of a small change in any one of their states. What’s even worse is that these are only those factors seen overtly. The model has no clue about what is going inside the person’s head. That is how difficult it is to understand human behaviour.
The Model lacks a Midbrain. But that does not have to mean that you can’t have an expert in the field intervene. The experts come in as replacement Midbrains to feed data that can help fill in and/or rectify gaps in information adding to the continued learning of the ever-powerful model. You are only as good as the company you keep. When the model is backed by ML masterminds and those that have mastered the workings of their own minds, (metacognition), the model continues to get better and better at developing an intuition of its own.
It is a tedious but ongoing task. The lobes have had a very fruitful and conditioned ‘give and take’ relationship that has worked through the ages towards rational decision making. This is exactly what we want the model to replicate allowing the ML engineers and Behavioural scientists to work as the 4 lobules and the midbrain.
Everyone wanted the forebrain, no one thought of the midbrain, until now…
If you liked reading this and would like to see more engaging and informative content, please visit iSchoolConnect or click on this link to see what our Masterminds have to say.
