In Between. AI and the science of silence and white space.

Published by

In Between Words

The science of the unspoken.

Soooo much texting. So much work focused on Natural Language Processing, and Speech Recognition...

I'll like to highlight some important work where Artificial Intelligence is focusing on the space between words, to do something good.

To be precise, not just the space between words but also other non-verbal patterns like the tone and pitch of a voice, or the time gap between the communicators in a dialog.

It all started in 1999 when researchers at the MIT Human Dynamics Lab proved the presence and power of social signals in human communication, and the ability of machines to detect and understand them.

Fast forward to 2007 when a startup named Cogito started testing an artificial intelligence platform, along with behavioral models, to interpret human communication and detect psychological states.

Cogito’s technology was quickly deployed in healthcare management programs to gather valuable data. Thousands of interactions were analyzed, generating millions of data points to better understand language patterns and behavioral models.

The AI system behind it is actually language agnostic: it does not care which language you speak. It doesn't care much about words and their meaning. The system looks at subtle cues in speech pattern. The Machine Learning algorithm focuses on the silence, energy, strength, frequency of interruptions, pacing, hesitations, tone, and the participation patterns of the speakers.

It's communications syntax analytics at its core. Non-verbal, non-gesticular, pure. It's already helping in crisis management, in children developmental analysis, in diagnosing post-traumatic stress disorder (PTSD).


In Between Marks

The science of the unmarked.

Can you please draw a clock? For many years now the clock-drawing test has been performed to diagnose, among other things, dementia and Alzheimer.

The following is from a 1998 article on the subject: "The clock-drawing test is a good screening test for dementia and cognitive dysfunction: normal clock-drawing ability reasonably excludes cognitive impairment. It is easy to administer, is not threatening to the patient, takes very little time, is easy to document graphically in clinical records and can be used to document deterioration over time." (1)

Now we have the DCTclock, an ingenious AI-driven system that looks for patterns that indicate early stages of cognitive impairment previously gone undetected.

Picture a Wacom tablet (or similar) and a digital stylus pen. The DCTclock AI records and analyzes all the performance data: the pressure of the stylus, the time between a mark and a subsequent one, the thickness of a line.

The new system focuses deeply on the absence of marks, evaluating the white space and the subtle hesitations in the drawing process.

With the new system, we are not just analyzing the final drawing, but rather the very process of marking to detect early sign of dementia.

By analyzing both, the process and the output, the system has the ability to see through compensatory strategies and detect cognitive indicators invisible to other testing methods.

AI is going to make us humans smarter. We have new ways to measure ourselves.


(1) BERIT AGRELL, OVE DEHLJN Geriatric Section, Department of Internal Medicine, Lund University Hospital, 22185 Lund, Sweden


« Back to articles