[ CogSci Summaries home | UP | email ]

Galantucci, B. (2005). An Experimental Study of the Emergence of Human Communication Systems, Cognitive Science, 2005, vol. 29, 737-767.

  author  = {Galantucci, Bruno},
  title   = {An Experimental Study of the Emergence of Human Communication Systems},
  journal = {Cognitive Science},
  year    = {2005},
  volume  = {29},
  pages   = {737--767}

Author of the summary: Sarah Lacelle, 2012, sarah.lacelle@gmail.com

Cite this paper for:


Human communication used today developed through combined cognitive abilities over generations into current shared behaviours. Most experimental research on human communication systems either focuses on methods using natural language, or using an artificial language designed by the investigator. These methods allow for investigation of acquisition, use and change over time of the communication system in question.

This experiment instead focuses on the emergence of novel communication systems in a controlled setting, without speaking or writing involved. Its purpose is to identify the cognitive processes used to create these new methods of communication.


Experimenters cultivated the emergence of a new communication system through use of a co-operative game scenario. Two players interacted in a virtual environment by each controlling an agent from different locations. Neither player was able to speak or see the other, and neither knew the other's identity. Communication was only permitted through basic agent behaviour and graphical communication that prevented the use of most numeral and pictoral representations. Success through three different game scenarios depended on cooperation among the players, generation of a method of communicating ideas and/or locations to each other, and further development of that system over the increasing difficulty demands of the games.

Players could communicate through means of a small digital pad and stylus. Symbols drawn on the pad were transmitted to both players immediately, but faded quickly. The horizontal motion of the stylus on the pad was preserved, however the vertical component of the drawings had a constant downward drift, regardless of the stylus movement. These properties allowed for three fundamental factors to be demonstrated: 1) it reproduced fundamental properties of spoken languages; fades quickly and interpretations are not straightforward, 2) prevented use of common symbols, and 3) the signal properties of the stylus and pad (frequency, location on panel, etc.) allowed for a wide variety of signs to be produced.

The experiment contained three different game scenarios. Pairs that successfully completed one scenario were invited to attempt the following game.

Game 1

|   |   |
| 1   2 |
|_ _|_ _|
|   |   |
| 3   4 |
Cooperative task taking place in a series of 4 virtual rooms. Each room wass identified with a unique symbol. At the beginning of each round, each player wass randomly located in separate rooms. The task required both players to arrive in the same room by each completing a maximum of a single move per round. Between rounds, players were able to move freely in the space until they chose to reset to a new round. Success in a round on a chance basis was 50%.

Players began the game with a score of 50 points, decreasing by 1 point per minute during the active part of each round. Failing to complete the task decreased the score by 4 points, and success in the task increased the score by 2 points. If the score reached 0 it did not decrease further. Game winning goal was to reach 100 points.

Game 2

|   |   |   |
| 1   2   5 |
|_ _|_ _|_ _|
|   |   |   |
| 3   4   6 |
|_ _|_ _|_ _|
|   |   |   |
| 7   8   9 |
Game took place in a series of 9 virtual rooms, including the same 4 as present in Game 1. All rooms were uniquely identified with a symbol. One room contained a prey. Prey was considered caught when both players were present in the same room as the prey. Capture of the prey reset it to a new room, but the game was continuous with no interim time as in Game 1


The score for this game also decreased over time, as well as decreasing if both players were found in the same room without the prey being present. Score increased only when the prey was caught. Goal of the game was to reach 100 points before the end of the session.

Game 3

|   |   |   |   |
| 1   2   5   10|
|_ _|_ _|_ _|_ _|
|   |   |   |   |
| 3   4   6   11|
|_ _|_ _|_ _|_ _|
|   |   |   |   |
| 7   8   9   12|
|_ _|_ _|_ _|_ _|
|   |   |   |   |
|13   14  15  16|
Game 3 took place in a series of 16 virtual rooms, containing the previous 9 as seen in Game 2. All rooms were uniquely identified with a symbol. Game 3 maintained the mechanic of capturing the prey, but introduced the element of two enemies. Found in two separate and random locations, each enemy only attacked a single player and ignored the other. When a player entered a room that contained their target enemy, the enemy began chasing that player through the series of rooms. The chase ended when both players met in the same room, resetting the enemy to another random location.

The score for this game also decreased over time. Having either player being chased by the enemy increased the rate of point loss. Score also decreased if players met in a room not accompanied by either prey or an enemy. The only increase in points came from catching the prey. Goal of the game was to reach 100 points before the end of the session.


Ten pairs or participants were recruited to complete Game 1. Players were not informed of the layout of the rooms, nor the transformations affecting the output of the digital pad.

After players reached a score of 100 points in the game, most pairs completing Game 1 in ~3 hours, players were asked to complete a separate session to asses the communication developed between the players. The test session was 10 minutes; 5 minutes without the communication system developed, and 5 minutes with it. The difference in score between the separate 5 minute sessions estimated the benefits of communication. The same assessment period took place after successful completion of Games 2 and 3 as well.

One pair was not able to solve Game 1, likely because one of the players' inconsistent use of given symbols for several different meanings. The same player made no effort to stabilize communication with the other player. Another pair was terminated from the study due to repeated cuing from the experimenter. For the pairs that successfully completed Game 1, the recorded scores accurately reflected how early a consistent communication system arose. The fall in score early in the game indicated when no communication was used, and a sharp rise in score was noted once pairs started to develop methods of communicating via the pad and stylus.

The study continued with the 8 successful pairs that were not terminated. Games 2 and 3 were designed to see how the pairs would further develop their established sign systems with more factors to account for. All pairs solved Game 2 in ~1 hour with no significant decrease in score. Pairs spent less time establishing a method of communicating. Game 3 was significantly more difficult, with two pairs unable to successfully complete it in the allotted session time. Game 3 required more sophisticated communication systems and optimized player behaviour than the other games.


Two major processes underlie the creation of communication systems:

  1. Learning-by-Using -- Player A draws an S while in an unknown room, and is currently uninterpretable by Player B. Player B then decides to move to Room 2 and finds Player A. Player B can then conclude that S stands for Room 2.
  2. Naming Procedures -- While players are in the same room, Player A produces a S while walking into the symbol identifier for that room. Both players can interpret that S would stand for that room. Players who used this procedure often suspended new rounds early in Game 1 to go around and create symbols to identify all rooms.

Successful pairs in Game 1 developed three distinct types of sign systems:

  1. Numeration Type -- Sequences of lines corresponding to an arbitrary number scheme for the rooms.
  2. Icon Type -- Various icons corresponding to specific rooms.
  3. Map Type -- Symbols used to indicate location in regards to the layout of the rooms on a map. Systems used by players in Game 1 indicated longitude only.

Signs often incorporated task-related information. Player behaviour often depended on information integrated with a sign, such as information about a player's location and environmental information. Time elapsed between sign creation was also relevant to the information they were meant to convey.

It can be concluded from the results of Game 1 that communication systems emerge fairly quickly and reliably, even by use of methods with many restrictions, making normal communication difficult. Also, effective communication can be had by different types of signs, depending on how and what they encode, as seen by most pairs developping different symbols for the same elements.

As pairs continued to develop their communication systems over the rest of the study, most pairs were able to solve Game 2 without much communication amongst each other. More overt communication was required to complete Game 3 however. Sign systems established in Game 1 primarily developed by the same method they were created, be it learning-by-using or by overt naming. During this development stage, sign systems that were similar among multiple pairs began to diverge and become more unique in their use of signs. As well, the same symbol within a single system could have multiple meanings encoded to it, if used consistently in this manner among the players.

Successful completion of both Games 2 and 3 relied not just on further development of sign systems, but also effective and efficient behaviour coordinating between players (i.e. a split-search strategy might be used for locating the prey).

It was shown that once a sign system had been established new signs were usually related in some way to the previously established ones. As to the relative shapes of the symbols, most systems were comprised of signs of dots or lines. Signs of this form are simple to visually distinguish and simple to create, and can accommodate individual variation.

Summary author's notes:

Back to the Cognitive Science Summaries homepage
Cognitive Science Summaries Webmaster:
JimDavies (jim@jimdavies.org)