Verified by 
Imagine a pair of scissors. Even if you have never seen or used them, immediately after you look at them, you know that the number of possible actions is limited. The holes are there so that you can put something into them, and the only logical thing that will fit are fingers. The holes are affordances: they allow the fingers to be inserted. The different sizes of the holes provide constraints to limit the number of fingers that will fit: the big hole suggests several fingers, the small hole only one. The mapping between holes and their movements and what happens in the world is clear and results in an immediate understanding, easily learned and always remembered. A pair of scissors is an example of good design and usability; its visible structure provides clear clues to how it works through its affordances, constraints, and mappings. However, the world is populated with poorly designed objects, which can be difficult and frustrating to use. They provide either no clues or, sometimes, misleading ones. The consequence is a world filled with frustration, with objects that cannot be understood and interfaces that lead to error. But having in mind the behavioral principles of affordance, constraint, and mapping can help us to shape a world with objects that are gentler to ourselves.
Affordances — The Psychology of Materials
Harry Harlow was an American psychologist best known for his maternal-separation experiments on rhesus monkeys, which manifested the importance of caregiving and companionship to social and cognitive development. How did Harlow go about constructing what was later called the “science of love”? He separated infant monkeys from their mothers a few hours after birth, then arranged for the young animals to be “raised” by two kinds of surrogate monkey mother machines, both equipped to dispense milk. One mother was made out of bare wire mesh. The other was a wire mother covered with soft terry cloth. Harlow’s first observation was that monkeys who had a choice of mothers spent far more time clinging to the terry cloth surrogates, even when their physical nourishment came from bottles mounted on the bare wire mothers.
Harlow’s work provided experimental evidence for prioritizing psychological over biological parenthood, while underlining the developmental risks of adopting children beyond infancy. In a much less considered perspective, it also provided us with a lot of evidence about the psychology of materials, as well as the affordances of two types of materials as different as terry cloth and bare wire. The mothers made of terry cloth satisfied the psychological needs of the infant monkeys that the bare wire mothers did not, even when they were the ones feeding the infant monkeys' biological needs. Although Harlow’s experiment was not intended to show that materials and things have a psychology of their own, it contributed to unravel the affordance of objects: the perceived and actual properties that determine just how a thing could possibly be used. Bare wire mothers cannot fulfill emotional and psychological nourishment. A chair affords (i.e., is for) support and, therefore, affords sitting. A cup affords us to carry and drink liquids, a pen to write letters and draw pictures. Wood is normally used for solidity, opacity, support, or carving. Flat, porous, smooth surfaces are for writing on. Glass is for seeing through and for breaking. Affordances provide strong clues for the operation of things. Knobs are for turning. Slots are for inserting things into. When affordances are put to proper use, the user knows what to do just by looking — no pictures, labels, or instructions are required. Complex objects require explanations, but most things should be simple. If simple things need pictures, labels, or instructions, the design has failed.
Constraints — Knowledge in the World
Most of the time, shopping malls have huge parking lots, where customers can conveniently park their car before going shopping or to the movie theater. To enter the parking lot, customers need to get a ticket, which they use to pay for parking according to the amount of time their car is in the lot. After that, they can insert the ticket into a machine that will open a gate, allowing them to get out of the parking lot. However, because the card is only readable on one side, most customers fail to figure out immediately which of the sides is the correct one and have to resort to a trial-and-error method, which causes frustration, wastes time, and increases traffic lines to leave the lot. This happens because there are no visible cues on the card (i.e., there is no knowledge in the world), and most of the time, there is no knowledge in the head of the customer, who forgot or did not notice which side of the card is readable when they took it out of the machine for the first time upon entering the parking lot.
Not all of the knowledge necessary for precise behavior needs to be in a person's head. It can be distributed — some in the head, some in the world, and some in the constraints of the world. Behavior is determined by combining the information in your memory (i.e., in the head) with that in the world. However, the world imposes constraints to allow behavior, because the physical properties of objects — the order in which parts can go together and the ways in which an object can be moved, picked up, or otherwise manipulated — constrain possible operations. Each object has physical features — projections, depressions, screw threads, appendages, etc. — that limit its relationships to other objects, the operations that can be performed to it, what can be attached to it, and so on.
The Physical Constraints of a Lego Motorcycle
Imagine a child who dismantled her Lego Motorcycle. How much does she have to remember to put the parts together again in the proper order? If there are 13 parts, there are 13! (10 factorial: 10 x 9 x 8…) different ways in which to reassemble them — a little over 6.2 billion alternatives. However, not all possible orderings can be produced: there will be physical, semantic, and cultural constraints on the ordering. Physical limitations constrain possible operations: a large peg cannot fit into a small hole; the motorcycle windshield would fit in only one place, with only one orientation; etc. Semantic constraints rely upon the meaning of the situation to control the set of possible actions. For the motorcycle, there is only one meaningful location for the rider, who must sit facing forward. The purpose of the windshield is to protect the rider’s face, so it must be in front of the rider. Semantic constraints rely upon our knowledge of the situation and the world, and they can be powerful and important clues.
Cultural constraints rely upon accepted cultural conventions, even if they do not affect the physical or semantic operation of the device. Cultural constraints determine the locations of the three lights, which are otherwise physically interchangeable. Red is the culturally defined standard for a stop light, which is placed in the rear. White or yellow (in Europe) are the standard colors for headlights, which go in front. And a police vehicle often has a blue flashing light on top. Logical constraints dictate, in the case of the motorcycle, that all the pieces should be used, with no gaps in the final product. People could use cultural constraints to figure out that the red was the stop light and should go in the rear, that the yellow was the headlight and should go in front, but what about the blue? For people who had no semantic or cultural information to help them figure out where to place the blue light, logic provided the answer — only one piece left, only one place to go. The blue light was logically constrained. Natural mappings, the next behavioral principle we will explore, work by providing logical constraints to the user.
Mappings — The Relationship Between Controls and Results
A natural mapping takes advantage of the spatial side of the functional layout of components and the things they affect or are affected by. If two switches control two lights, the left switch should work the left light, the right switch the right light. If the lights are mounted one way and the switches another, the natural mapping is destroyed. Some natural mappings can be cultural and biological, such as the universal standard that a rising level represents more, a diminishing level less. Similarly, a louder sound can mean a greater amount. Amount and loudness (and weight, line length, brightness, and even sensitivity to money) are additive dimensions: we need to add more to obtain the same incremental increase. Note that the logically plausible relationship between musical pitch and amount does not work: Would a higher pitch mean less or more of something? Pitch (as well as taste, color, location, and smell) is a substitutive dimension: you need to substitute one value for the other to make a change. There is no natural concept of more or less when we compare these qualities. Other natural mappings can follow from the principles of perception and allow for the natural grouping or patterning of controls and feedback (see Figure 3).
Figure 3 — Seat Adjustment As a Perfect Natural Mapping
Natural mappings have the power to reduce the need for information in memory. Think about the arrangement of burners and controls on a standard kitchen stove with four burners arranged in the traditional rectangle. If the four controls are arbitrarily arranged, the user would have to learn each control separately: that's 24 possible arrangements. Why 24? If the first control could work with any of the four burners, that would leave three possibilities for the next leftmost. So there are 12 (4 x 3) possible arrangements of the first two controls: four for the first, three for the second. The third control could work either of the two remaining burners, and then there is only one burner left for the last control. This accounts for 24 possible mappings between the controls and burners (4 x 3 x 2 x 1 = 24). With a completely arbitrary arrangement, the stove is unworkable, unless each control is fully labeled to indicate which burner it controls. However, the use of spatial analogies can relieve the memory burden. A common partial mapping that is in frequent use today is to segregate into left and right halves. That will leave to the user the task of “only” needing to figure out which left burner each of the two left controls affects, and which right burner each right control affects — that's two alternatives for each of the four burners. The number of possible arrangements is now only four — two possibilities for each side — quite a reduction from the 24. But the controls must still be labeled, which indicates that the mapping is imperfect. Since some of the information is now in the spatial arrangement (i.e., its knowledge is in the world), each control need only be labeled back or front; the left and right labels are no longer needed. Still, we can make the system even more user-centered, by using a proper, full natural mapping, with the controls spatially arranged in the same pattern as the burners. This will allow the organization of the controls to carry all the information required. We know immediately which control goes with which burner. That is the power of a natural mapping: the number of possible sequences was reduced from 24 to one (Figure 4).
Figure 4 - Arbitrary (a) versus Full Natural (b) Mapping
Conclusions — User-Centered Design
Applying these behavioral principles will allow companies to design for usability and understanding and to discover a new competitive edge, enabling their customers to save time and money, while increasing morale. The main principles to achieve a good user-centered design are: 1) use constraints to make it easy to determine what actions are possible at any moment; 2) make things visible, including the conceptual model of the system, the alternative actions, and the results of the actions; 3) make it easy to evaluate the current state of the system; and 4) follow the natural mappings between intentions and the required actions, between actions and the resulting effect, and between the information that is visible and the interpretation of the system state. In other words, make sure that: 1) the user can figure out what to do; and 2) the user can tell what is going on.
Design should be able to understand the natural workings of people and of the world; it should exploit natural relationships and natural constraints. As much as possible, it should operate without instructions or labels. If explanations are necessary — and especially if the explanations lead the user to think or say, “How am I going to remember that?” — then the designer has failed.
References
Norman, Donald (1988). The Design of Everyday Things. New York: Basic Books.
