The joy of “anticipation” (Part 1 of 2)

Simulation Game

Every single moment, our brain runs through hundreds of simulations. With that, it aims to reduce surprises and guarantee our survival.

Most of them happen quickly, effectively and unconsciously.

Welcome to anticipation mechanisms!

Before we get into some more details, I want to quickly differentiate “anticipation” and “expectation”. Let us open the Cambridge Dictionary (CD) and have a look:

“Expectation” reflects certainty of something to happen. “I expect the train to leave in the next 5 minutes.”

Expectations leave little options. When it doesn’t happen, the surprise is big(ger).

Anticipation allows for flexibility.

So what? Anticipation mode on…

Robert Rosen focused the majority of his life researching anticipatory systems, he defined anticipation as:

To paraphrase: anticipatory systems are models (it could be human, animal, AI), that use knowledge about their current state, and their environment to get into a specific future state, after reviewing multiple future state options.

Anticipation is one of the oldest and most important mechanisms in living creatures.

This process can happen multiple times, within seconds or over a longer period of time taking into consideration multiple inputs sources.

For example, when you place your feet on uneven ground while on a hike and your brain anticipates the degree and form of the ground under your shoes. Or when birds anticipate winter to come and migrate to warmer climates. Or when you execute a project and utilize governance and Worst-Case-Scenarios to mitigate errors.

Let us have a look at different anticipatory capabilities in animals:

If you care to read more about anticipatory system in other organisms, this article provides a nice deep dive into “what” and “why”.

Past and Present

The “cost” of surprise can be positive and negative. The difference between what the brain anticipated based on experience (top-down) vs. what is actually happening at this point in time (bottom-up, in-the-moment sensory input, e.g. see, hear, feel, smell, touch) is called “prediction error”.

A positive prediction error (things turned out better than anticipated), will be stored in our brain as “enjoyable” and “to be repeated”.

A negative prediction error (things turned out worse than anticipated), will be stored as “unpleasant” and “do not repeat”.

Both recalibrate existing knowledge, every single awake second throughout life.

This is the underlying mechanism for learning.

From picking up that first spoon, that you probably shoved into your eye or ear (just ask your parents), all the way through to that first large product release, you and your experienced team managed just fine.

Top to bottom: what is real?

Another fascinating effect of “top-down” (experience) and “bottom-up”, is that “bottom up” perception is impacted by “top-down” knowledge. This is called hierarchical predictive coding of anticipatory processes.

So, let me give you four examples to make it more vivid, because this is important to realize:

1. Bird or a bat?

Last week, I went for a run at dusk. I saw a big, black bird against the pink-blue evening sky. That’s what I thought I saw. Only it was not a bird, but a big black bat. We don’t have big bats in Germany. My brain anticipated a bird, because it only knew big black birds.

2. Pea or wasabi?

My daughter picked up a pea from the floor after dinner. Only it was not a pea from her dinner plate a minute ago, but a wasabi nut from the evening with friends the night before. She anticipated a fresh pea and got a spicy surprise. She was not happy!

3. In our Out?

Ever seen a professional tennis game (table tennis or squash?)? These players run towards the ball almost at the same time their opponent fires a shot. After years of training, the players’ brain anticipates where the ball could end up in just milliseconds with pretty high precision. An amateur would still stand around and think, while the ball ends up behind her.

4. Signed or not signed?

A US Sales Director flew to India to close a big deal for the first time. All signs showed green lights. During the trip, the potential client was warm, welcoming and spent two days with the Director, good talks and nice small talk during dinner. Based on his US-focused experience, his “gut feeling” was positive. Only the signature did not come through. His Indian counterparts did not find him vested enough. His own beliefs and experiences overrode slight nuances, he could have caught and which would have helped him to work harder. Culture played a game on him.

Recent studies showed, that our brain filters out nuances, we (well, the brain) might “think” are unimportant. The brain can also “fake” or gravely impact incoming visual signals, because it can become so “sure” the bat is a bird, that it “sees” a bird not a bat. Or it can “tune” certain signals up or down, so we do not pay attention to them at all, but only to what our own experience has taught us to be relevant.

These are all very helpful mechanisms, especially trying to survive in the wild forest 100,000 years ago. But some of them might get into our way during modern era life.

And action

Last but not least, it is important to connect anticipation with action. When we anticipate certain things to happen, we get ready: emotionally or physically. When our brain accepts the reward potential as big enough (vs. risk) and hence worth spending energy on, dopamine comes into play and gets us moving. (We will look into dopamine soon as well!)

Anticipatory processes happen all the time. Without anticipation, we would not be able to walk, socialize or invest.

Our brain goes through anticipatory scenarios to select the action that is most likely, and combines the highest rewards with the lowest risk. This is not always optimal for our modern lives. The brain likes to stay in a safe harbor.

Anticipation is largely based on prior experience and beliefs. The impact of “top down” experiences on “bottom-up” sensory inputs creates our very “own” reality and risks missing important cues (e.g. within cultural environments).

Next week, we will look into the impact of anticipatory processes in social and cultural settings. This comes in handy when you work with different people from different backgrounds. We will also look briefly into anticipatory systems within AI / NLP research, as models often fall back on similar anticipation logics.

Let me wrap up with a quote:

Thanks for reading.

Alex