Monkeys, Bananas, and Data Science

A scientist puts a single piece of lettuce in a cup and tells a monkey to come for his meal. The scientist repeats this experiment every day. Monkey comes in, finds cup, eats lettuce. Soon enough, this monkey is bored.

Until one day the scientist puts a banana in the cup. Monkey comes in, finds cup, expects lettuce — and instead finds a banana. This monkey is ecstatic.

This is how monkeys (and humans) learn. It is also how machines learn. At Apttus Accelerate in San Francisco, cognitive neuroscientist Dr. Carmen Simon discussed the connections between these human behaviors — including these experiments — and machine learning in business technology. Dr. Simon focused on three points: our use of mental schemas, the role of expectations in forming habits, and the importance of surprise.

Drawing Mental Schemas

As humans, we write our own scripts every time we decide and remember how to act in different situations. It saves us energy — and as a result, things become familiar: going to the grocery store, driving to work. Familiarity becomes predictability, and predictability allows us to determine what will happen next.

But it’s a fine line: when stimuli start to become too familiar, your brain starts to habituate to them. You no longer notice the interior of the supermarket or the road signs on your drive.

Creating Expectations

The more we are exposed to something, the more we form expectations of that thing. Our brains map out everything we do: events unfold in your brain before they take place in reality. We need to feed the brain’s constant craving for what happens next, so we try to predict those outcomes. When we’re used to a certain outcome, we learn to expect it without thinking twice.

Introducing Surprise

So what happens if you don’t give the monkey the banana? The scientist starts to mix it up: sometimes the monkey gets lettuce, sometimes bananas. When you don’t present a reward every time, but only half of the time, you’ve created uncertainty, which is the key element of surprise. (Not to mention, a potentially furious monkey.)

When you find a gap between what happens in your mind and what happens in reality, the brain learns. It looks at the two together and becomes a more powerful prediction engine. Machines learn in the same way. For people and systems alike, learning from experience becomes the most trustworthy way to improve.

When you find a gap between what happens in your mind and what happens in reality, the brain learns. It looks at the two together and becomes a more powerful prediction engine. Machines learn in the same way.

Dr. Simon’s point is an extremely valuable one for enterprise technology. We aren’t just users of enterprise software, robotic potential buyers and opportunity owners.

We’re humans.

So how can we create technology that learns — that is more human? Dr. Simon highlighted a few questions that help us focus on anticipation, uncertainty, and rewards in the user experience:

• As you improve your systems, have you built a technology that the user will find rewarding?
• Do you provide your users a framework that will make them feel familiar — without feeling restrained?
• How can you match your users’ level of satisfaction and then change it up to introduce surprise?

Any successful enterprise software company has to ask these questions. We interact with tools the same way we interact with each other: constantly improving and learning from one another. Dr. Simon drew real parallels between human and machine learning, and how advanced data science can drastically improve enterprise technology, especially in sales and customer relationships. At Clari, our team has thought a lot about how we use systems as people, where gut instincts end and the analytics begin. The line isn’t always clear.

Sometimes that’s for the better. It means the two are seamlessly woven together.