Keynote: “Good Enough” – The Myth of 70% Emotive Accuracy by Dan Hill, CEO, SensoryLogic #ISC2015 #MRX
- Self-report has its limitations
- In the New York Times, men report 1.6 billion condom uses per year but only 400 million are sold annually
- 95% of mental activity is subconscious
- It’s not just lying, it’s what people can’t tell you
- Self-reported information can be lip service, filtered input
- Emotional metrics include recall, call to action, preference, satisfaction, loyalty
- We have a sensory brain, emotional brain, and rational brain
- Emotional brain is 10 times more active , emotional responses have 5 times more quickly than cognitive responses
- Consumers are well informed but not rational well informed
- 20% of adults in the western world are functionally illiterate and have no real command of the English language. [and yet we write the longest most complicated surveys and expect everyone to answer them perfectly]
- Dial data – there is a time delay, do we assume that if someone doesn’t move the dial it is the same affect as the last time the dial was moved – “padding” the data
- Once dial data gets to a certain level, it doesn’t change even though people might be really bored
- It’s difficult to get someone to understand something when his salary depends on not understanding it – think TV ratings
- The last five seconds of a tv spot is tough territory because people know they’re getting a sales pitch at that point. It needs a really clever hook.
- Facial coding vs self-report correlated 12% for cognitively filtered feelings
- Athletes show a lot of fear, they remember the fear of losing. Do guys self-report fear? Very rarely, I’m tough, I’m a road warrior.
- EEG measures cognitive load, attention, and arousal. Correlates highly with IQ.
- [so conclusion, every research method has flaws.]
- Recall the TV show “Lie to me” – scientist could tell when people were lying even if they didn’t speak
- Charles Darwin new that emotions were universal, spontaneous, abundant – blind people emote the same way as everyone else, it is hardwired. Social display rules vary – expressions may be shorter or longer. Otherwise same by gender, religion, culture, etc.
- Dan wrote “Emotionomics” – for the business investor
- Automated coding is still probably only 34% accurate. Automation over reports because of random noise
- Unless they’re faking it, full emoting of happiness doesn’t last longer than 4 seconds.
- Software can’t pick up micro-expressions – like when you’re bored with someone but you’re still smiling
- Tilting your head can change the automated scoring completely unrelated to the emotion
- When the mouth moves as you talk, reliability of software declines
- 70% is perceived as commercially viable since 90% sounds like a lie and 60% sounds bad. But the software just isn’t there yet.
- Linking verbatims to emotions provides solid insights – you might say something but not really mean it – reasons aren’t causes
- There is no ‘lying’ muscle in the face J
- Could not code Michael Jackson because of too much surgery [sounds like a way to get around facial coding!]
- Automation is prevalent now but it is buyer beware
Live blogging from Nashville. Any errors or bad jokes are my own.
– are you a purist or pragmatist about quality? [i like to think i’m both. you know what’s right. do it right. i don’t see why that’s so difficult]
– why do we need a revolution? maybe self-report doesn’t always work. men report using 1.6 billion condoms, women 1.1 billion condoms per year. but companies only sell 600 million per year. [hmmmm… who to believe :)]
– 95% of our behaviour is subconscious. how are surveys doing in this area?
– 95% of water below the waterline of the titanic is what sank it
– not everyone can be honest with themselves, not everyone understands why they do things, not everyone wants to be honest with you or themselves about why they do things
– you cannot think your feelings, you have to feel your feelings
– self-reported information is always filtered
– emotional metrics are subconscious – recall, call to action, preference, satisfaction, loyalty
– it’s not simply left brain, right brain. the emotional brain sends 10 times as much to the other side as it receives
– you can’t rewire the human brain. we are closer to homer simpson than Spock
– you can ASK when people are engaged during a commercial or you can measure each second
– in a commercial, people turn off in the last 5 seconds when it turns to the sell job [pay attention conference presenters]
– when people are asked to move a dial to show how they’re feeling, they forget to move the dial. that’s not very accurate
– people are particularly bad at sharing opinions about negative feelings.
– charles darwin did work on emotions – emotions are universal, even blind people across cultures has the same facial expressions, face is spontaneous and fast as the muscles are attached directly to the skin, we have more facial muscles than anywhere else on the body
– theory refined by Paul Ekman, matched emotions to specific muscles
– believes facial coding in 3 years will be good enough and ready to go [i’ll take that bet 🙂 ]
– fMRI brain scans were the closest to their manual facial coding. automated facial coding isn’t there yet
– 40% of time during commercials have no reportable emotion. really, it’s not as interesting as the birth of your child.
– best joke of all time, you might laugh for 4 seconds
– political campaigns are the energizer bunny of fake social smiles
– people will say they are happiest for 12 seconds but that’s just not reality
– surprise and anxiety often go together
– need self report, we need all the data we can get and self report is about as good as other data right now
– there are many things that people won’t tell you, perhaps you can see it in their face