I am not talking about real dead elephants. I have never, in fact, found a real dead elephant. I am talking about an odd experiment suggested to me by a research assistant some years ago. This was a visual search experiment. You were either looking for a dead (well, inverted) elephant among live (upright) ones - like this.

 

 

 

 

 

 

 

Or you were looking for an upright elephant among the inverted - like this

 

 

 

 

 

 

In an experiment like this, we vary the number of elephants (known in the trade as the "set size") and we measure the amount of time it takes you to say "yes", the target elephant is present or "no" it isn't (the "reaction time"). What we found was that the reaction time increased as the set size increased, but it increased faster when you looked for "dead" among "live" than when you looked for "live" among dead. Your intuition may or may not be congruent with that result. After all, these pictures are just single examples. Observers in the experiment would have done this task for hundreds of trials and they would have had a somewhat harder time finding the dead elephant.

Why?

And, perhaps you may ask, why would we care?

I am interested in visual search. You spend a significant amount of time looking for objects even if they are literally right in front of your eyes. So, for example, look for the letter "T" in this collection of Ls. It isn't hard to tell the difference between a T and an L but, in order to find your target, you are going to have to search from L to L to L until you stumble on the T. If you think you found the T instantly, you probably just got lucky. Go find the second T. There are two in there.

You need to search because your brain simply cannot do everything everywhere at the same time. So you cannot "read" all the green letters at once. Indeed, you are probably limited to recognizing one (or may a very few) objects at one time. You may think that you are simultaneously recognizing many objects in the scene before you right now, but it isn't so. That is an illusion (and a topic for another day).

You can't do everything everywhere but you can do some things everywhere. So, in this picture, you can find the red T very quickly because you can appreciate some aspects of color all over the visual field at once. Consequently, you can guide your attention to the few red things without have to waste time and energy on the green items.

Search is important. You go to the buffet. Where are the plates? Where are the forks, the salad, the potatoes, and so on. You sit down. Where is the fork now? Where is the salt? You need that fork again. You don't notice the endless collection of searches because most of them begin and end so rapidly that they don't draw your attention. It is worth noting that this raises an interesting point for a column on "attention". Attention isn't A Thing. It is a name for a large family of processes involved in selecting some aspects of the input or of your internal states for further processing. You use a form of visual selective attention to do these mundane - find the fork - search tasks. It is perfectly possible to say that you perform these acts of attention without paying attention to them. Two different types of attention are involved.

Other search tasks can be far more important and nowhere near as easy as finding that fork. Consider searching x-rays of luggage for threats or searching medical images for signs of cancer. So, one might imagine, there would be some benefit to understanding how we search; how we get our attention onto the current object of interest. Fine, but what do dead elephants have to do with this?

One of the interesting features of search tasks is that they can be dramatically asymmetric. Search for Thing A among a bunch of Bs might be much easier than search for B among A. Some of these asymmetries make a degree of sense. As Anne Treisman reported 30 years ago, it is easier to find the presence of something than to find its absence. So it is easy to find one moving dot among stationary ones. It is harder to find one stationary dot among moving ones. Other cases are a bit trickier to explain.

Why is it easier to find the tilted lines among vertical distractors on the left side of this picture and harder to find the vertical lines among tilted lines on the right? Maybe vertical is the absence of tilt and, like the absence of motion, harder to detect.

This doesn't do much for the elephant asymmetry unless you think that ‘deadness' is the feature. Some researchers have suggested that the more deviant or novel item is easier to find. Thus, a mirror-reversed letter "N" can be easier to find among "N"s than an N among mirror "N"s. The inverted elephant might be more novel than the upright elephant. However, when we tried the same experiment with swans, we didn't find the asymmetry. It was just as hard to find an inverted swan as an upright one. The asymmetry held for camels: Dead camel - easy, live camel - hard. The rule governing this behavior is not clear.

Searching for silhouettes of animals is not really a major piece of our research, but the elephants have nagged at the back of my mind for a decade. Given a chance to write this blog, I thought I would take the opportunity to present the puzzle to a new set of eyes and brains. Why does that inverted elephant seem more salient than its upright brethren? Your turn.