Computer-based Incentive Models for Individuals (hint, Amazon, Google, Facebook)

The purpose of this blog is to present an outline for how a neural net computer system might model the neurological phenomena associated with incentives.

Why? What is the value of this line of investigation? The answer is at the end of this brief introduction about why artificial intelligence (AI) is such a big deal to business.

As society embarks on neural nets and AI, one of the area’s that online business is deeply invested in is its customer behavior patterns. Understanding the online customer and their behavior patterns enables online business to provide improved customer experiences. The less friction between a customer and their (the customer) desired outcome, the reason for their online presence at the business site, the higher the potential for customer satisfaction, return visits, more purchases, etc. Let’s put the scientific proof of that previous statement to the side for the sake of this articles investigation.

Hopefully we can agree that online companies, like Google, Amazon, and Facebook, collect a ton of data related to their customer online behaviors. Additionally, hopefully we can agree that these companies will use artificial intelligence to analyze that data to look for behavior patterns that will enable the customer data owner to provide improved customer services.

In the previous two statements we established:

• Some large online entities will collect data representing their customer activities

• Some large online entities will use AI to extract information to help them provide improved customer experiences

Without betraying personally identifiable information (PII), like name, address, etc. these large online entities are able to collect individual entity behavior patterns, by tying the behavior data to single key items, things like source IP address plus machine and browser type, or whatever combination of data they can combine to create what would be a unique identifier for that returning customer, no PII. This is an important distinction because at this point, the large online entity is tracking individual behavior without associating said individual with a name, address, or any other PII, yet, they are collecting data that represents individual behavior patterns. They key here is data that represents individual behavior patterns.

Why is data that represents individual behavior patterns important?

Because, and this gets back to the title of this blog, if it were possible to build computer models that:

1. Approximates an individuals behavior patterns

2. Would be able to incentivize the specific entity to behave in a certain way, say, buy a specific item

3. And these two models worked together

The result would be a computer-based incentive model for individual identities.

The business result: a highly targeted and very highly accurate computer-based incentive system aimed at individual online customers (individuals based on data representing behavior patterns gathered per online identity).

The neuroscience of incentives – the foundation of a computer model:

“Incentive salience is a percept-bound type of “wanting”, which typically occurs as relatively brief peaks upon encountering a reward or a physical reminder of the reward (a cue). Incentive salience does not require a clear cognition of what is wanted, and does not even need to be consciously experienced as a feeling of wanting, at least in some cases (though when it is brought into consciousness it can considerably intensify feelings of desire). “

Wanting and Liking: Observations from the Neuroscience and Psychology Laboratory

What is the value of this line of investigation?

If you buy into the idea that collecting data about online customer behavior has a huge business value, and if you buy into the idea that neuroscience pretty well understands how incentives work in humans, then it would seem reasonable that we could build computer models of incentive systems. If, say, Amazon, were to utilize an online entity behavior pattern, and if Amazon wanted to position goods and services that would have a very high likelihood of being selected for purchase by that entity, they could consider building a computer model for incentives, based on the neuroscience of incentives. This would save Amazon a ton of money. Yes, right now they probably look at your online patterns to guess what you might want next, given what you have done in the past. What do you think their return rate is on the guess? Probably pretty high based on their history of data for your online activity. What if they could improve that rate by orders of magnitude by leveraging targeted incentives based on a model of your mind – a model refined by the data that represents your online activity. Amazon could save a boatload of money it currently spends presenting guesses, albeit highly refined guesses based on historical trend data. Instead of having to spend money associated with the processing and online presentation of guesses, let's say, and this is just a guess, 1 out of every 100 presentations turns into a purchase, with single identity targeted incentives modeled after the very real neuroscience of incentive behavior, the odds of success just skyrocketed, enabling Amazon to pocket (or not spend) what they would have spent doing the guessing game presentations.

Now, this just establishes a business case for:

1. Building a behavior pattern computer model of a targeted online identity

2. The value of building a computer model for incentivizing online identities

Once something is established, and found to work, there typically follows a refinement period – a period where the system that has found a level of success has attention given to it to make it work even better. Look at Microsoft Windows. It is still in its refinement phase. The point is, once online incentives start, and are targeted to individual identities, consider where the refinement phase will lead.

If I can get to you behave in a highly predictable fashion, I now have control over you. <cue song for twilight zone>

Read the draft of Book 1: Ten @ https://sites.google.com/view/time-a-trilogy/

Twitter at @hankmgreene or https://twitter.com/hankmgreene

Facebook: https://www.facebook.com/hankmgreene

Flipboard: HankMGreene

Notes:

1. Dopamine is the main system for reward. Dopamine particles are actually released before the reward, hence thought to support the importance of environmental stimulus and increasing reward seeking (and being released before the reward is a big deal) A computer model would need to mimic this behavior relative to a specific identity in order to understand how to incentivize the identity. This is not an impossible task.

2. Nucleus accumbens (an area to strongly understand as an influence for the incentive model) is strongly associated with motivation and reward. This is the area to build a computer model for in order to create an effective incentive system.

3. Mesocortical dopamine travels via the frontal cortex (the focus of wanting) – part of the reward system. Additional parameters for the computer model.

References:

The Neuroscience of Natural Rewards: Relevance to Addictive Drugs

http://www.jneurosci.org/content/22/9/3306

6 Key Ideas Behind Theories of Motivation

https://www.verywell.com/theories-of-motivation-2795720

Wanting and Liking: Observations from the Neuroscience and Psychology Laboratory

This is a fantastic article that drills into the details of the neuroscience of want – a means to structure impactful individual incentive programs

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2813042/#!po=3.48837

Wikipedia: Mesolimbic pathway

“The mesolimbic pathway regulates incentive salience, motivation…”

https://en.wikipedia.org/wiki/Mesolimbic_pathway

Great 2 minute YouTube video

https://www.youtube.com/watch?v=f7E0mTJQ2KM