Research finds split in perception of similarity that could double Web advertising

The University of Warwick research, is published this month in the journal Cognition and was led by Dr Zachary Estes, found that people differ radically in their perceptions of similarity. Whereas some people base their similarity judgments on physical features that looked alike, others based their judgments on more theme based relations. For instance, some people thought a bee is more similar to a butterfly, whereas others thought a bee is more similar to honey.

Dr Zachary Estes said: "Similarity underlies many of our behaviours, such as where we look for coffee in a supermarket and what social groups we belong to."

The phenomenon we have studied has many potential applications - a clever supermarket might, for instance, sell up to twice as much of one type of coffee by placing it both with other coffees and also with biscuits. However this could have even more impact on some forms of web based advertising such as Google AdWords. Some advertisers may for instance have selected words with physical similarities to their product such as cake and cookie but ignored large sections of their audience who would make a strong connection between cake and birthday".

Dr Estes and his team were surprised at how stark the split was between a theme based and physical based choice and even more surprised that the split was so even.

His first experiment using the word "similar" found that 31% of the people chose a physical feature based similarity (for instance cake and cookie) whereas 46% went for a theme based similarity (such as cake and birthday). The researchers tried changing the word "similar" to "like" to try and rule out any strange semantic effect but this time 31 % still selected a physical based likeness (such as fur and hair) whereas this time 57% selected a themed likeness (such as fur and coat).

A further experiment asked the participants to consider both similarities and differences at the same time. This changed the balance somewhat with 62% selecting a physical feature based similarity and 25% a theme based similarity.

This later experiment gave Dr Estes a possible insight into why and how there was such a radical split in these similarity judgments. The third experiment pushed people into thinking a little more about both similarly and difference. Could it therefore be that those people who were quick to make to a decision plumped for "easy" themed linkages (such as crown and Queen) whereas those who took more time to think about things chose a more considered physical link (such as crown and hat)?

Further experiments indeed showed that there was a group of people who rushed into a decision and tended to chose a theme based similarity and that there was a second group of people that thought about things a little more chose a physical feature based similarity.

However what was most surprising was that the Warwick researchers found a third group of people that also thought about things a little more and yet still always chose a theme based similarity. That left an overall split in the population of around 50% who were more thoughtful and chose a physical feature based similarity (such as net and rope) and another 50% for who always went for the thematic option (net and fish) whether they used instinct or deeper thought.

Source: University of Warwick