1. Thinking about Thinking
and the demands of the post-digital world
Cassandra Moore, PhD
Aquilent
IA Summit 2014
2. The digital revolution is over.
Nicholas Negroponte
We have a growing internet of things.
Data is ubiquitous, sensors are autonomous.
Post-digital is the state of things after the fanfare.
How do we dwell in an information ecology?
@Seymour188
6. How do we build digital systems to
resonate with those parameters?
@Seymour188
Everything we think, feel, sense is
filtered through the particular
parameters of our sensory and
cognitive systems.
7. Two Families of Mental Operation
System 1 System 2
• Automatic, effortless, quick
• Largely unconscious,
difficult/impossible to control
• Operates on biases and heuristics
• Does most of the work of thinking
• Controlled, effortful, slow
• Usually conscious
• Logically coherent, rule-governed
• Lazy, difficult to engage
• Who we think we are
@Seymour188
8. Paying more When Watched
Bateson, M et al, Cues of being watched enhance cooperation in a real-world setting. Biology Letters (2006) 2, 412-414
15. Associative Coherence
• Suppression of ambiguity or doubt
• Incorporation of extraneous information
• World seems much more coherent than it is
@Seymour188
20. Substituting Similarity for Probability
Linda studied philosophy in college , most people think she went to Berkeley. She
participated in anti-nuclear marches, she was very bright and very active.
Ten years have passed. Which of the following is most probable?
Linda is a feminist bank teller.
Linda is a bank teller.
P(Teller) P(Feminist and Teller)
@Seymour188
21. We Suck at Statistics
• Substitute easier operations for hard ones
@Seymour188
• Discount the base rate
• Small numbers are lawless
22. It’s tough to make predictions,
especially about the future.
– Yogi Berra
@Seymour188
23. Impact Bias
• Overestimate the
impact of events on
future happiness
• Focus solely on the
event, tend to forget
other things will occur
• Best predictor is how
others are doing in the
same situation
24. Overconfidence Bias
• Pundits’ predictions
around chance
• Overconfidence in
predictions
• Discount falsifying
evidence
• Reluctant to change in
the face of negative
evidence
@Seymour188
25. Where do we go from here?
• Expect everything to become a coherent story
• Multi-tasking isn’t
• Under-design – causality comes for free
• Tell people’s stories to improve decisions
• Do the stats behind the scenes
@Seymour188
28. Percentages are Difficult
A health survey was
conducted on a sample of
adult males of all ages and
occupations.
What percentage of the
men have had a heart
attack?
What percentage of the
men are both over 55 and
have had a heart attack?
A health survey was
conducted on a sample of
100 adult males of all ages
and occupations.
How many of the 100 men
have had a heart attack?
How many of the men are
both over 55 and have had a
heart attack?
Hinweis der Redaktion
The digital revolution is over. Which is not to say that we’re no longer interested in the digital realm, or that we’ve gone on to something else. It has become less remarkable, but no less pervasive.
We have an internet of things, we have symbiotic learning systems that we really can’t say we control.
I think of the post-digital as a term analogous to post-punk, or post-feminist. Punk and feminism haven’t gone away, they still significantly affect our culture, but the shifts they produce are more subtle, mutations rather than disruptive changes.
Our challenge now is to really live in the world of information. To live well.
In order to do this, I think we must also talk about the “we” that are living in the in the information environment.
So l invite you to spend a little with me thinking about thinking
In the Clock of the Long Now, Stewart Brand suggests that society is also constructed of layers that change over time at different rates of speed. The bottommost layers are slower to change, they are more stable. The top layers change quickly and provide for innovation.
He situates technology in the commerce and infrastructure levels. Somewhere toward the top.
In contrast, we humans, being part of the nature layer, are much slower to change.
Let’s think about that for a moment.
The evolution of the human (and most other) species is very slow.
>Anatomically modern humans – appeared approximately 200 thousand years old on the plains of Africa
>We were mostly concerned with avoiding predation, feeding ourselves and getting our genes into the next generation.
>Complex speech production only happened 50 thousand years ago
>The first visual designers appeared about 40K years ago
>And the first written content only about 7K years ago – not sure when content strategy appeared.
Dates of course, are subject of debate but you get the idea. We took our time getting to the modern era.
During the vast majority of this time our bodies, brains and behavior were dedicated to survival.
Thinking had to be fast, based on little information, and very predictive.
Probably not a lot of difference between thinking and perceiving.
Certainly not a lot of contemplation and reasoning. Plato had to wait.
During that time, we evolved cognitive/perceptual systems that resonate with the environment in which we lived.
We walk erect, light comes from above, things are on the ground below, and the horizon is in the distance.
Our systems have particular characteristics that were probably selected for by evolution
For example, we neither perceive ultra-violet nor infra-red light as some species do. In fact “visible light” constitutes a very small part of the electromagnetic spectrum.
What would it be like to see like a bee or a snake or a bat? It’s hard to imagine. They have a different view of the world than we do.
Yet for us, with our particular lenses, it’s difficult to believe that we don’t open our eyes and perceive the world as it is.
I’m going to suggest that everything we think, feel, sense is filtered through the particular parameters of our sensory and cognitive systems.
The information environment is what we apprehend, not what’s out there.
Our cognitive and perceptual systems evolved in synchrony with the physical world.
Now, we are challenged to build a digital world that resonate with our natural cognitive and perceptual systems.
I suggest we start by taking a look at those natural systems.
Two systems approach – two general families or categories of thinking.
Which is not to say that you have two little homunculi in your head – but by naming them it’s easier to think about their characteristics
System 1 is the older system, perfected for escaping predation, surviving in the Serengeti
System 2 – cognitive strain is type 2 kicking in
Michael Gazzaniga, the neuroscientist, says System 2 is who we think we are.
I am going to concentrate on System 1, because it is our default approach to thinking and most likely the mode of operation of most users when interacting with the information structures we’ve created.
Melissa Bateson, University of Newcastle, conducted an elegant experiment that illustrates the operation of System 1, the quick, largely unconscious family of thought.
As in many office kitchens, the psych department has a donations box for tea, coffee, and milk and a list of suggested prices. At the communal beverage stand, the scientists posted changing images close to the coin catcher, alternating photos of human eyes with pictures of flowers.
People put nearly three times as much money into an ‘honesty box’ when they were being watched by a pair of eyes on a poster, compared with a poster that featured an image of flowers.”
Unconscious behavior adjustment – they were not aware of the significance of the changing posters.
It is a bit disturbing that we can be so unaware of influences on our behavior, or our motivations for doing particular things. But these “illusions” about our thinking are not negative.
System 1 also enables people to act intuitively, based on habit or expertise. When there’s no time for reflection, or no perceived need for it, system 1 takes the lead.
Understanding the two systems and how they work can lead us to better decisions about architecting information such that it resonates with our natural ways of thinking.
When we need build a rapport with Homer rather than Spock
How we present (and represent) information to make it easily consumable, while ensuring that the message is received.
I’m going to talk about a few of the characteristic biases we observe with System 1. There are many others and I encourage you to look into them further.
Associative coherence refers to our tendency to make whatever is happening into a coherent story.
Jorge Arango has written at in the Journal of Information Architecture about links and nodes in information structure.
Our associative network of concepts can described as links and nodes. Concepts are nodes that activate or prime other concepts via links.
Priming prepares us to interpret what comes next. It activates nodes.
Seeing “Vomit” activates the connections you have to it in associative memory, temporarily dampening the activation of other concepts. You would be faster to recognize the words “hangover” and “sickness”.
Instead, you see bananas. You cannot help but associate those two concepts. If you are like most people, you made a causal story connecting the two words, and chances are that bananas won’t seem really appetizing right now.
Another example of priming. Participants in this experiment were asked to read the last four digits of the serial number, then respond to the question “When did Attila the Hun rise to power?”
The higher the number, the more recent Attila’s rein. The number is totally extraneous, but it gets incorporated into the story of Attila the Hun because associative coherence, story-making, is a characteristic of intuitive thinking.
Anchoring
You may have heard of “change blindness”
We’re blind to the obvious and blind to the blindness
Two teams passing basketballs, task is to count passes, thousands miss the gorilla. Without the task no on misses the gorilla
Without specific attention to the relevant object, we don’t see it. Attention is allocated elsewhere, the gorilla does not fit the story of passing a basketball, so it’s presence is ignored.
****
Creating a coherent story includes ignoring information.
The invisible gorilla has obvious implications for Google Glass and heads up displays in cars. If the task of the display user demands attention, driving might not be so safe.
**Bring in bullets one at a time
Adopting one interpretation and rejecting another.
If extraneous information has focus, it will be incorporated into the story. Anchoring (was $79.99, now only $49.99), knowing prices of medical procedures
http://youtu.be/sZBKer6PMtM
This is a video of 2 triangles, a disk and a square moving around on the screen.
Causality is inferred because of the timing of the intersection of motions – the triangle touches the square, a part of the square moves.
The intent to harm the little triangle is attributed to the large triangle.
As the shapes continue to move around the screen, personality characteristics are attributed to them – the big triangle is usually seen as a bully. The circle is timid.
Even infants see causality in movies like these and are surprised when the expected behavior doesn’t happen.
Notice that this is intuitive thinking bordering on perception.
Timing is everything.
Agency is attributed to objects
Motives and personality are attributed to objects.
These are characteristics that we can use to better convey a message. Just by characterizing two patterns of thought as System 1 and System 2, Homer and Spock, we make them more memorable.
The attributions made by intuitive thinking come along for free.
Ever taken a stats class? I taught stats to psychology students. They hated me.
Probability and statistics are hard – you really have to think about them – engage System 2.
But system 2 is difficult to engage.
Seeing similarity, however is easy – it’s intuitive.
Since our System 1 is doing most of the work, and System 2 really doesn’t want to be bothered, we substitute the easy operation for the difficult one.
Here’s a demonstration.
Linda studied philosophy in college , most people think she went to Berkeley. She participated in anti-nuclear marches, she was very bright and very active.
Ten years have passed. Which of the following is most probable?
First option: Linda is a bank teller
Second option: Linda is a feminist bank teller.
Most people say the second option is more likely.
But note, the probability of being a teller is greater than the probability of being both. The base rate – the number of tellers – is greater than the number of feminist tellers.
But being a feminist is associated with the story we’ve formed about Linda. Probability is hard, so we go with similarity instead.
Law of large numbers – extremes are much higher in small samples, change over time is more likely
It’s tough to make predictions, especially about the future.
Dan Gilbert, in Stumbling upon Happiness, reports when we think about how events will affect us in the future, we overestimate their impact.
We imagine being happy for the rest of our lives if we win the lottery, forever depressed by the loss of a limb.
Focus solely on the event, forget that there will be many other events occurring as well. Other events dilute the impact of the big event.
But billionaires are not radically happier than you and I, and amputees are not necessarily depressed.
The implications of the impact bias for someone working in health care are huge. How do we give people faced with treatment options the information that will help them make a decision they’ll be happy with? We tell them other patients’ stories.
Just as with viewing a visual illusion, knowing it’s an illusion won’t make it go away, but it can help us make decisions that will make us happier.
Around chance when more than a year out.
Say they’re 80 – 90% confident, when they’re 60 – 70% correct
We focus on information that supports our beliefs and ignore the rest
Our inner Homers discount falsifying information, we really don’t want to change our views
Series in NYT on the confidence of Donald Rumsfeld – utter certainty based upon gut feelings
Contrast to Nate Silver. Using System 2 -- applying statistical analysis rather than heuristics to polls. His accuracy rate is much higher.
Does your head hurt? Is System 2 tired? We’ve been through a lot of pointy-headed stuff in a very short time. If you’re still with me thank you.
So let’s just wrap up here.
Expect everything to become a coherent story. Review your site/app as a whole and see what story it’s telling. Look for anchors that might be helpful to getting your point across, make sure you don’t have random serial numbers.
We cannot successfully multi-task, we switch between tasks and each one takes a performance hit. If an error message is important, give it a window with focus. Make it modal if it’s really important.
As Stephen Anderson was saying yesterday in talking about sandboxes – under-design – causality, intentionality, personality all come for free. Design is not decoration – it influences the story that’s being told.
Help people make better decisions - whether it’s about buying a chair or choosing a medical treatment
Do the stats behind the scenes analyze the data from the sensors and provide information we can intuitively understand. And do your own stats – calculate how many participants you need to get statistical significance, analyze your date – it will make your findings so much more convincing.
And different species experience the world in different ways, depending upon the systems they have evolved.
Raptors have two foveae in each eye.
We have stereoscopic vision because both our eyes are on the front of our heads. Unlike the raptor whose field of view is much wider than ours because its eyes are on the sides of their heads. Also unlike the raptor, we do not have natural visual processing and motor control that would enable us to fly at speed through a forest chasing prey without running into trees.
We don’t fly, we walk, so our systems are adapted to the being terrestrial and moving at lesser speeds.
****find the content that goes directly to make the point and trim away what’s not important. Highlight “percentages”
Typically, the percentage of men both over 55 and having a heart attack is thought to be higher than the percentage who have just had a heart attack. People say the second is more probable than the first.
When asked the number in 100, however, people do not overestimate the dual probability.
May be easier to create a spatial representation of number and reason more easily about it.
P(A), P(A and B), not P(A|B)