Haptic emotional communication

How much information does haptic quality add to communication?

A hand conveys very rich emotional information. There are 5 fingers, each able to convey data such as pressure, curvature, temperature, rotation and movement. Note the picture of a child holding onto an adult finger. We often see similar scenes when children are being led somewhere by an adult.It interesting to ask what the essence of the emotional information was:

What is the smallest amount of information we could have in a haptic signal that could convey emotion?

Related Research

A review was done on existing haptic emotional communication devices.The following three devices were most relevant:

InTouch explores emotional communication by mapping touch to rotation.

LumiTouch explores emotional communication between couples by mapping touch to light color and intensity.

Sentograph, by Manfred Clynes, is a 3D pressure transducer that relates emotional content of music to haptic pressure. He found that emotion and pressure reaction to music was universal.

ComTouch: A one-finger device

The ComTouch pad device maps squeeze force to vibration on one finger. There is one input channel-- a force sensitive resistor (FSR) at the fingertip. When the FSR is pressed, the force is translated to vibration intensity. There is a local feedback area to feel the vibration from the sensor is being pressed, and this vibration is also sent to the ComTouch of a corresponding user.

Experiment Design

The task is to tell if the picture one person is seeing can be identified using only ComTouch to communicate emotion.

Hypothesis:One-finger of communication can improve emotional communication over straight guessing.

Here are some experimental design decisions made with help from the researchers in Affective Computing Group:

The design of the experiment must have a control group to establish a baseline.

· A ground truth is necessary—a way of measuring whether a goal was achieved makes the hypothesis easier to prove.

· As there are many uncontrollable and undesired factors in the lab setting, simplifying the experiment is desired. This resulted in the elimination of many variables (e.g. telephone, color photos).

· Invented a control experiment.

Experimental Method

Each test involves viewing sets of pictures. The experiment was conducted on pairs of friends, one is the sender, and the other is the receiver. The sender receives a book of pictures, while the receiver has a contact sheet containing all the pictures. The sender uncovers the pictures one at a time and tries to communicate emotion perceived. When the receiver has an idea of which picture is being viewed, she writes the order of the picture on the contact sheet. No talking except the vocal cues of “next” or “repeat” are allowed. When the task is finished asked them what method they used.

To determine the baseline communication between the subjects, a control experiment is first conducted without the device; only vocal cues of “next” or “repeat” are used as the subject flips through pictures. Next, a brief explanation of the device is given, and 5-10 minutes are given for familiarizing with the device. The experiment continues with them repeating the photograph experiment, except this time using the ComTouch to try to convey emotional feeling from sender to receiver.

The experiment was performed on seven pairs of subjects. The control test consisted of 3 photographs, while the ComTouch test consisted of 7 photographs.

Recording Observations

From the discussion, users could be separated by their method of communication in two groups: Emoters and Coders:

Coders used a method designed for photo content. Emoters used a method that relied on a mix their own impression in addition to a guess of their partner’s emotive reaction to the photo content. Furthermore, emoters could describe their mapping generally relating pressure to the intensity of feeling, while duration was related to the intensity of activity in the picture. However, no consistent mapping was found, among the emoters in their assignment of the pressure or vibration.

Results & Analysis

On the whole, the added channel resulted in slightly better performance, as 4 out of 7 pairs did better than chance.

Comparison of different schemes

The methods used to convey the emotional quality varied in that 4 used a coding scheme while 3 used emotion. Coders did not benefit as much as emoters benefited from single channel of finger information.

Conclusion

One finger of information seemed to have an impact. People did say they could identify some emotion, but some of the pictures were too ambiguous. The study was inconclusive as there was a large error, and the sample size was too small.

Evaluation

I had not planned to use this device for the experiment, so am surprised to have gotten relatively good results. This experiment was low resolution (not enough data sets) with high error. It was surprising to me that people with emotive intention fared better.

Coders might have done worse due to human error (e.g. forgetting the code) or inability of the established code to adapt to the images presented.

Future Work

This preliminary study will need more subjects and a larger sample size of pictures in order to prove anything conclusively. Some subjects also suggested using more highly arousing pictures or more pictures, as most of the pictures used displayed low arousal, positive valence affect.

Thank you

(Subjects) Noga Livnat, Zahra Kanji, Joe Stark, Caroline Cutting, Karolina Netolicka, Vishy Venugopalan , Rory of Fifth East, Jim Gouldstone ,Goce Zojcheski, Landi Parish, Vimal Bhalodia , Dave Wilson, Ian Finn, Sarah Farrar, Rob Jacob and Gian Pangaro.

(math) Abinhav Kumar

(experiment design) Roz Picard, Raul Fernandez, Carson Reynolds, Ashish Kapoor

Go back. Worksheet of Experimental Data Anjchang HomePage Anjchang MAS630 Page Go forward.