Current Issue Past Issues About YSM Subscriptions Advertisements Contact Us
78.2 - Winter 2004
From the editor
Departmental notes
In Memoriam
Where are they now?
Undergrad profile
Book reviews
Web search
25, 50, 100 years ago
Puzzle corner

Search YSM Articles
Advanced Search

> Winter 2004 > News

Using Robots to End AI: Artificial Inference
Printable Version

By Hannah Yoon

Nico, an upper-torso humanoid robot, is designed to socially interact with children.
Nico, an upper-torso humanoid robot, is designed to socially interact with children. (Credit: Andrew Lovett)

Andrew Lovett ’03 and Assistant Professor of Computer Science Brian Scassellati, recipients of the Best Paper Award for “Using a Robot to Reexamine Looking Time Experiments” at the third International Conference on Development and Learning (ICDL) in San Diego, CA, used robotic experiments to validate an alternative theory of why infants spend longer times looking at certain objects.

The researchers showed that infants do not look longer at a particular stimulus because children consciously decide to look at a surprising event. Nico, the model robot, played an instrumental role in this proof, allowing researchers to follow the computerized visual processing that occurred in Nico as the robot faced the same constraints and cues facing actual infants.

Sixteen computers control Nico, who observes the world through four small cameras that provide both wide and narrow fields of view. Nico’s detection of color, touch, and motion are incorporated into data used to direct its social behavior.

Scientists can test cognitive theories or inferences by reading how feedback from the real world interacts with a robot’s cognitive operations. Unlike a human clinician, Nico can also be controlled to interact objectively with real children, allowing scientists to observe how different children respond to Nico’s behavior. Lovett observed that many researchers at the ICDL shared a profound interest in creating models of the human cognitive process, in hopes of obtaining a better understanding of human brain development. “The ideal final product would be a humanoid robot capable of learning everything a human knows, and thus reaching human-level cognitive capabilities.”


Lovett, A. & Scassellati, B. (2004) Using a Robot to Reexamine Looking Time Experiments.  Retrieved November 30, 2004, from
Science Links

Copyright 2014 Yale Scientific Publications, Inc. - Disclaimer