Decision Mechanics

Insight. Applied.

The neuroscience behind bad decisions

By

People often make irrational decisions. An article in Quanta Magazine describes how too much choice can often be to blame.

The article also discusses the attempts of neuroscientists to understand the mechanics underlying irrational decisions. Technologies like functional MRI (fMRI) are driving the emerging field of neuroeconomics. Neuroscientist Paul Glimcher claims to be

…aiming for a grand unified theory of choice.

One theory coming out of this research is that divisive normalization plays a role in choice. Divisive normalization comes from research into the visual system.

The brain is a power-hungry organ—consuming 20% of our energy while representing only 2-3% of our mass. It therefore makes sense for us to have energy-conserving optimizations in our cognitive processes. The visual system removes information that is uninteresting and highlights the surprising and novel.

Some neuroscientists believe that a similar process is in play when we are confronted with choices. When presented with a choice between 2 or 3 items, the process copes well. However, when faced with the myriad choices that are characteristic of the modern consumer society, the process fails us and leads to irrational decisions.

The theories coming out of neuroeconomics are controversial. It’s a young field. However, using the tools of neuroscience in this way could certainly lead to fresh insights into longstanding questions in economics.

Glimcher says that the research has already improved his own ability to make good choices.

Rather than pick what I hope is the best, instead I now always start by eliminating the worst element from a choice set. […] I find that this really works, and it derives from our study of the math. Sometimes you learn something simple from the most complex stuff, and it really can improve your decision-making.

R Notebooks

By

RStudio have announced R Notebooks. This adds an interactive execution model to R Markdown documents.

In R Markdown documents text and code can be mixed to create active documents that perform and update analyses as they are run.

However, R Markdown documents are executed in batch mode. This makes them more suitable for reporting than for teaching and investigation.

With the introduction of R Notebooks, code segments in the R Markdown document can be edited and executed in isolation—making the whole experience much more interactive.

R Notebooks can be saved as text (HTML) files allowing them to be shared via GitHub, for example. The HTML files can be viewed in a standard web browser as static documents. However, when the files are opened in RStudio, the R Notebook is automatically recreated from the HTML.

Support is also provided for languages other than R, including Python, C and SQL.

For more information, check out the development team’s YouTube video.

AI-first world

By

In his recent keynote Google CEO Sundar Pichai said

In fact, when I look at where computing is headed, it’s clear to me that we are evolving from a mobile-first to an AI-first world.

Devices like Amazon’s Echo and Google Home are “always-on” intelligent assistants designed to sit quietly and await our bidding.

As the devices recede into the background it’s the AI-powered services that will come to the fore.

Machine learning at scale on HDInsight using Microsoft R Server at Spark

By

Microsoft have published an article on how to conduct a decision tree analysis using Microsoft R Server and Spark on Azure HDInsight.

Using four 8-core 28Gb RAM (D4) worker nodes they were able to process 170 million rows (37GB) in around 5 minutes. This was 20% faster than using Spark’s own MLLib libraries—although there’s no comparison with Spark’s newer ML libraries.

Latin American Leadership and Effective Governance seminar

By

Leadership and Effective Governance seminar poster

I gave a talk last month on interactive decision making at a Leadership and Effective Governance seminar run by George Washington University. The seminar is aimed at Latin American politicians and civil servants.

My talk covered topics such as

  • stakeholder analysis
  • game theory
  • confrontation analysis

The techniques were illustrated by examples of confrontations facing the new government of Argentina. I built an online immersive briefing describing the confrontational landscape facing President Mauricio Macri and his team.

However, given the pan-American immigration debate fueled by Donald Trump, much of this was applied work was (unsurprisingly) overshadowed by impromptu discussion and analysis of Brexit.