Like a map, a GIS is layer based. And like the thematic layers in a map, GIS datasets represent logical collections of individual features with their geographic locations and shapes as well as descriptive information about each feature stored as attributes.
Prior to GIS, mapmakers created a series of map layers that were used to geographically describe and characterize a location. They often used transparencies that could be overlaid on a light table. These integrated displays were used to visualize spatial relationships and gain insight about relevant characteristics of a place. Practitioners would use these to make interpretations and to draw interesting conclusions.
One visionary who used this process for planning was Dr. Ian McHarg, a landscape architect and renowned writer on regional planning using natural systems. His seminal book was published in 1969 and articulated concepts for ecological planning, which applied these map overlay principles. You can learn more at Wikipedia about Ian McHarg and his work.
Around this same time, Dr. Roger Tomlinson, known as the "father of GIS," developed his early ideas for GIS. Among other aspects of GIS, he further articulated the concept of thematic layers and overlays as a cornerstone for GIS.
These early GIS practitioners thought about how geographic information could be partitioned into a series of logical information layers—as more than a random collection of objects. They envisioned homogeneous collections of representations that could be managed as layers. These GIS users organized information in individual data themes that described the distribution of a phenomenon and how each theme should be portrayed across a geographic extent. They found that they could use relatively simple GIS data types (points, lines, polygons, and rasters). These simple data layers could be combined through location—that is, georeferencing enabled datasets to be combined in a map or overlaid using geoprocessing operations such as polygon overlay.
These pioneers also provided a protocol for data collection and how to manage these collections as geographic data layers. Here is one example for representing soils.
Each and every area (polygon) in a specified extent could be assigned a dominant soil type, and the soil types could be consistently classified and described using properties or attributes of each polygon. In the case of soils, very involved sets of properties are typically recorded for each soil polygon.
A theme could be defined to delineate various areas representing the dominant soil type (that is, a layer collection of soil type polygons and their descriptions as attribute values).
This organizing principle of geographic layers became one of the universal GIS principles that provided the foundation for how GIS systems represent, operate on, manage, and apply geographic information.Source: http://help.arcgis.com