Mood Brightener: ...more from Stay Homas. (Confination IX)

 

Models:

A model in the context of science is description of how we think/understand some process takes place. Models can be purely descriptive or explanatory or some blend. An important feature of all models is that they evolve based on a process that is some form of the one illustrated below. This itself is a model for how to describe the scientific process. While it looks a little intimidating if you read each step and think about your personal model of relationships you may find that it matches pretty well (we all have some kind of model of relationships!) This description of the scientific process comes from one of my mentors Eugenia Etkina and is called the ISLE model.

© 2015 ISLE Physics Network. All rights reserved. Used by permission.

What it boils down to is observe, notice patterns, create a trial explanation, and predict the outcome of a test of your explanation. If your explanation seems to work in this case we test it some more in other contexts and if it seems to work consistently we use it to guide the development of our relationships. If the initial, or some subsequent, prediction is inconsistent with the results of a test we then consider what might be wrong with our model and modify it accordingly. We then use our modified model to make further predictions and test them and on and on it goes. Some models are relatively simple and others are tremendously complex. A model is only as good as the tests we subject it to and our ability to discern whether our prediction agreed with the results of a test.

Consider the prosthetic leg (Herman Herr) you learned about a while back. In order to construct a leg that could be trained to respond like a real dancer's leg the team had create a model (mental model at first) of what they thought a leg was doing. I'm sure there were many surprises and false starts as their model succeeded and failed are various tests. Like the rest of our human experience I'm sure it succeeded at a number of tests and then failed to perform well in some unpredicted way. As Tim Minchin would point out this process is always changing the model incrementally and is never done. At some point it's good enough or as good as we can do and our development of the model pauses until some future opportunity.

Theo Jansen: Strandbeests

Here's an interesting example of a endlessly developing model from a computer scientist turned kinetic artist.

What would you say is Theo's model for a strandbeest? Has he gone around the testing and adapting cycle a couple of times or many many times?

If you get fascinated by this stuff here is the link to his videos where he explains some details of how these creatures were created. He actually used an evolutionary model in a computer to arrive at the design! He also explains the neurons and sensors he has developed for these creatures. In the context of this class it is an interesting question whether these are robots or machines.

My more general point is that Theo had an initial model for how this might work and has repeatedly traveled around the scientific process curve for many years improving the original model and creating spin offs that continue to fascinate nearly everyone.

COVID:

Since we've already started down the road of applying many of the science concepts we learn here to our current COVID situation I feel like there is real value in continuing. Our models about how the SARS-CoV-2 virus is transmitted and how the illness (COVID) created by contracting the virus progresses are actually are actually lovely examples of how this model building scientific process works and the confusions many folks have about that process. As I'm sure you know it's called a 'novel corona virus' because it's new to us. This explains why our initial models for this corona virus started with what we know about other corona viruses. What we have seen over the months since January is a consistent cycle of predicting what we would expect to happen IF our current model is correct. Then that model is tested as we apply that model to our best attempts to help those who are infected. Two specific examples are perhaps useful to explore.

Corona Viruses Go Away with Warmer Weather:

The flu is a corona virus and it does go away with warmer weather. There are a number of possible reasons for this but this reality is why we have a 'flu season' that starts sometime in the fall and finishes in the late spring. It's not that nobody gets the flu in the summer but it's much less likely. There are also viruses that do NOT go away with warmer weather and while our initial model was that SARS-CoV-2 might go away we needed to test it. While many of the initial outbreaks were in places that were experiencing winter and early spring temperatures there were also places in the southern hemisphere and the tropics that were significantly warmer. What have we learned in the intervening months about this feature of our model for the SARS-CoV-2 virus? Does this mean the initial model was wrong for the SARS-CoV-2 virus? Did scientists say that they thought it would go away with warm weather? Did the scientists follow the appropriate process for the 'incremental acquisition of knowledge'?

Note from Class: Several people discussed the idea of the credibility of sources in evaluating whether to accept data or statements. A resource I mentioned and shared was this assessment/plot of the bias and reliability of various news sources. I provide it here for your future reference.

Cloth Masks Do or Don't Help:

I know this is a hot button issue but let's try to stay focused on the process of model building that we are discussing. In contexts where we have philosophical 'skin in the game' it can be hard to separate those perspectives from the science ones which makes this discussion good practice for me as well as you. No doubt you remember that there was some early discussion back in February and early March about the potential value of masks for the general public given that we were donating all our N95 masks to front line health care workers. Our original model came from a previous SARS virus that was very deadly (15% mortality roughly) but very difficult to 'catch'. Most importantly, perhaps, only people who were showing symptoms were infectious! For this previous virus there was no particular benefit outside of health care settings in broad public wearing of masks because one had to be in very close contact with an infected person to catch it yourself. This contributed to early statements that masks might not be important. Many of these corona viruses are spread by respiratory droplets and N95 mask definitely make a difference. In some cultures wearing good masks during flu season has been shown to reduce infection. In the begining we only had the model of previous viruses and we were very unclear whether this virus was easy or difficult to 'catch' though there were some indications that it was relatively easy to get infected. In previous studies of the standard flu it was found that if people stayed 6 feet apart the likelihood of infection was reduced. This is the root of the social distancing recommendation but it's based on a different virus. As the 'tests' came in it became clear that this corona virus is easier to catch than the previous SARS virus and some asymptomatic people were passing on the virus. Both of these bits of information required significant modification of the model. This led to the changing recommendations for mask wearing. Wearing a simple cloth mask helps reduce the size and range of the infectious droplets an infected person spreads. Masks plus social distancing seems to reduce the transmission of the virus. Was the original model wrong? Did the science community say that the orginal model was correct? Is the current model correct with certainty? Is the current model 'better' than the first model? Is 6 feet (2 m) far enough apart? (new data suggests maybe not)

In each case the science has been proceeding normally which is to say it keeps changing and generally keeps improving. What you do with that understanding is a different matter. If you want to dig deeper into these sorts of questions I would recommend the podcast 'This Podcast Will Kill You' which has done an excellent series about COVID-19 starting on March 23.

Autonomous Vehicles:

So here's an opportunity to explore this process of model building in a context that will eventually apply to our robot. Imagine you are driving up to a intersection that has a light. Those of you who drive do this regularly and you have a process you go through as you approach the light though you may be only subconsciously aware of it. Let's see what your model is for the process of approaching a light and deciding what to do.

Class Discussion:

A number of processes were discussed. All of them started with a dependence on whether the light as red or yellow or green.

If the light is red: In this case everybody agreed that there are a set of steps that we go through including taking your foot off the gas, checking in the rear view mirror to see if someone is close behind and might not notice you slowing down, checking ahead to see if there is a car in front of you, gauging the distance to the light to decide how hard to brake, and considering whether the light might turn green as I approach.

If the light turns green (from red) there are a number of additional checks we maker from assessing whether some numbskull is going to beat the red light on the cross street and will go through the intersection late, are there pedestrians still in the cross walk, are there bikers doing some unpredicatable bike thing, and what about deer?

If the light is green: One option is to keep doing what we are doing (moving smoothly) or we might choose to speed up if we think the light might soon turn red. While we are doing that we are also considering what we will do if the light turns yellow. We are also watching to see if other drivers are moving into turn lanes or slowing down.

If the light turns yellow (from green) there are a number of new assessments we make. Can we stop easily before the light turns red? Is there someone in front of me that might make a different decision than me? Do I need to accelerate to be through the intersection safely? What about all the biker/pedestrian/deer issues from up above.

If the light is yellow: Can I safely stop before I get to the light? If so slow down and stop otherwise continue through the light but watch in case a person stopped going the other direction suddenly tried to beat the light or there is a last second left turn in front of you. Is the person in front of me slowing down which trumps any other decision I might make?

What I hope you gathered from this discussion is that we all have very complex models for how to negotiate an intersection with a street light. When others described safety checks they do others were nodding along. Our next class we will discuss how we implement this model (an algorithm) and some basic coding tools for doing so.

Assignment Breadcrumb Reading: Bb Quiz

Models:

The quiz will ask you a number of questions about the characteristics of scientific models. These will be true/false or multiple choice and you will have multiple attempts so that we can solidfy our understanding of scientific models.

Before Next Class:

Assignment HW: Bb Quiz

SARS 2002 and Masks:

Based on our discussions and your reading (here is a reasonable source) why were masks not part of the recommended response for the general public? What feature of COVID-19 makes masks in public much more important for this corona virus?

Assignment HW: Bb Assignment

Updating Your Model:

Describe a model you have for how something works. This could be a model for a physical object (electromagnetic fields under power lines make you sick) or a behavioral model (if I do good things life will be fair to me). Describe how your model was tested and changed in some important way. You need to connect your model to a specific experience or event that led to a specific change.

Looking Ahead:

Look ahead to the next Breadcrumb:

Assignment Breadcrumb Reading: Bb Quiz

Decision Making:

What are the three primary decision processes described in this breadcrumb.