Jan 21 2014

A New Wrinkle on Change Blindness

I have discussed previously the phenomenon of change blindness – look at a picture which then winks off and then back on again. In between something may have changed. Would you detect it? Psychologist have found generally that people are pretty bad at detecting such changes.

Here is a nice demonstration of this phenomenon  by my colleague, Richard Wiseman. Just search for “change blindness” on Google and you will find many more.

Interestingly, if the change occurs without the picture winking off, in other words it occurs before our eyes, we are pretty good at detecting the change. Our attention is drawn to the change. But when the change occurs outside of our vision, we are bad at detecting that a change has occurred.

A similar phenomenon is inattentional blindness. Here is the classic demonstration of this.  (If you have ever seen it, take a look before reading on.)

The lesson with both of these related phenomena is that our brains process only a subset of the sensory information that comes in, and we attend to an even smaller subset.

How much information, however, is processed subconsciously – without our conscious attention?

A new study published in PLOS One takes a peek at this question. They conducted a classic change blindness experiment, showing subject pictures that wink off and then back on, and a change may or may not have occurred between the two. Researchers Howe and Webb asked subjects if any change occurred, and what was the change.

They found that subjects in most but not all of the trials were better able to detect that a change had occurred than to be able to name the specific change. They did the trial with faces, inverted faces, and simpler stimuli (green and red rectangles). The biggest difference occurred in the latter trial:

When a change occurred, observers detected the change on approximately 88.7 trials. Of these trials, observers went on to correctly identify one of the disks that had changed on approximately 64.2 trials. This meant that approximately 24.5 of the trials were only-sense, which was significantly more than the 3.2 trials that would have been predicted by a guessing strategy, t(9) = 8.99,p<0.001, Cohen d = 4.27.

So in this trial almost a quarter of the time (there were 100 trials total) the subjects were able to state that a change occurred without being able to identify a specific change. This was true of 13 of the subjects in the face trial, and 20 in the inverted faces trial.

In the rectangle trial discussed above the proportion of red to green rectangles changed. The researchers also ran a trial where red and green rectangles were swapped but the proportion did not change. In this trial there was no significant difference in “sense only” outcomes from chance.

What does all this mean? Some of the time we are able to detect that a change has occurred even if we are not aware of what that change is. This is likely due to processing involving the whole scene, rather than paying attention to and remembering specific details.

In the fourth trial where the rectangles changed but not the proportion, subjects could not detect change unless they were also able to identify a specific rectangle that changed. This makes sense because in this simple situation, only the details changed, not any overall aspect of the picture.

With faces, however, subjects were sometimes able to know that a picture had changed even when not aware that it was the glasses that had changed. This implies that our brains get an overall impression of a face (even an inverted face) and this overall impression is influenced by the details, even when we are not consciously aware of those details.

This is in line with previous psychological studies that indicate that subjects can react to cues of which they are not aware. We may respond emotionally to a face or be biased in our reactions, even when not aware of which facial features are generating that reaction.

This all fits in generally with how we currently understand human brain function – specifically that a great deal of processing occurs at a subconscious level and we are consciously aware of the net effect without being aware of all the elements that contributed to that net effect (how we feel, or what we decide, for example).

In this study there is also the specific element of visual processing. Our ability to detect that the proportion of green and red in a picture has changed, for example, may be more due to specific visual processing than something more general about attention.


As with most similar research, the message that the general public can take away from such research is to be aware of the limitations and foibles of how our brains function. Taking in and being aware of information is a process, and that process is flawed, incomplete, and biased.

The end result is that we can never be entirely sure of what we think we remember about what we think we experienced.

Ben Radford also discussed this experiment and made an interesting observation – that such subconscious awareness might give us some insight into why some people believe they have psychic powers. He argues that people have the ability to take in a great deal of social information about people without being able to necessarily identify which details influenced their assessment.

Alleged psychics may be fooled by this process into thinking that they are getting information from a paranormal or extrasensory source, when in fact they are simply subconsciously processing visual or other sensory information.

While I think this is essentially correct, I’m not sure this study directly relates to that conclusion. This study was about change blindness, and not about the subconscious processing of social cues, but the underlying principle, that of subconscious processing, is the same.

In any case – the study does contribute to the overall psychological and neuroscientific literature which clearly adds up to the conclusion that we all need to be aware of some of the basics of how our brains work, and this should make us humble in the face of the uncertainty this implies.

23 responses so far