My decades-long involvement with Anne Treisman’s Feature Integration Theory (FIT) must have begun in 1986. I was a junior faculty member at MIT at the time, working on visual aftereffects and binocular vision. I didn’t know much about attention. One might say that I had not paid attention to attention. Then, in 1986, Anne published an article in Scientific American (Treisman, 1986). I am not sure that I read her article when it came out, but my senior colleague, Whitman Richards did and he came to me to tell me that this was something I should look into. In my possibly dramatized recollection, I remembering him, striding down the hallway, waving the article, and declaring that I needed to do something about it.
The problem from our vision sciences vantage point had to do with the “feature” part of Feature Integration Theory. In a nutshell, Treisman’s FIT argued that there were a set of basic features that were processed in parallel across the visual field, and these features were not ‘bound’ to their objects. You might have green and round in the same area of the visual world, but to know that you had a Granny Smith apple, the green color and the round shape needed to be bound to each other. That binding (or ‘feature integration’) was a capacity-limited process that required that attention be directed to the object. Treisman thought that the basic features that were being bound were the same set of features that stimulated cells in the early stages of visual cortical processing. After all, there was evidence that separate extrastriate visual areas were processing different properties like color and motion (e.g. Zeki, 1978). This raised the problem; if motion was processed over here and color was processed over there, how did you come to see the moving red fire engine as a single, coherent thing? Attention, binding, and feature integration formed an answer to that question.
Looking at that Scientific American article, Richards thought that the mapping of Treisman’s “preattentive features” to the features of early cortical processing was too strong. I went on to do some experiments that we thought confirmed Whitman’s view. The ‘basic features’ are later representations of visual properties and do not reflect the first stages of cortical processing. Moreover, I and others began to question other key bits of FIT. A few years later, I offered up “Guided Search: An alternative to the Feature Integration model for visual search”. Mounds of other data appeared, challenging different aspects of FIT. Multiple other theories were offered. The details are not critical here (for one review, see Wolfe, 2014). For present purposes, the question is; “If FIT is wrong and 40 years old, why is there a two-volume special issue of Attention, Perception, and Psychophysics and why are we embarking on a week-long digital event, here, on the Psychonomic Society Featured Content site?
While it is true that subsequent data and theorizing contradict or complicate almost every bit of FIT, important theories retain power because they contain fundamental ideas that are correct. To briefly digress, this is why I taught Freud when I taught Intro Psych. The details of psychoanalytic theory are wrong, but core ideas like the idea that our motivations might not be entirely conscious, remain important. In the case of FIT, I can illustrate this point with this figure.
As you are searching around, you should notice a few things. Notice that, at first glance, you knew that the display was composed of red, blue, and yellow line segments, in pairs, and in a mixture of orientations. Notice that determining the identity of any specific item requires some sort of processing of just that item. At some point, you will notice that there is no upright, red and yellow T. I have been trying to figure out how you know when to stop searching for over 25 years. It turns out to be a tough problem. I promise that you will be more successful if you look for tilted blue and yellow Ls. So, it is obvious that something was visible as soon as you looked at the image. You knew about the colors, orientations, and something about the shapes “preattentively”. That preattentive knowledge is all you would need to confirm that there are no green frogs in this image, for example. Also obviously, you did not know that upright, red and yellow Ts were absent and that tilted blue and yellow Ls were present, until you started to search and to ‘bind’ the features together. That is the enduring core of FIT, the rest is commentary.
To provide some of that commentary, we (pictured below) have a week’s worth of posts, highlighting the reach and influence of FIT. Hayden Schill starts the week by looking at the parents of FIT. Where did these ideas come from? Sang Chul Chong discusses ensemble perception, that ability, mentioned above, to know something about the colors, orientations, and shapes in the figure above, before you know what specific items are present. Liqiang Huang highlights the search for new basic features. Nurit Gronau and Cathleen Moore will each discuss the nature of “objects”, including the phenomenon of “illusory conjunctions” where two features in the display get incorrectly bound together into the percept of an object that is not there. Finally, Ayşecan Boduroğlu considers how what we have learned about these topics can help us with a real-world problem; in this case, the creation of informative graphs. But wait, there’s more! We will have one additional post from Lisa Fazio, drawing your attention to the visual search capabilities (and scientific possibilities) of the archer fish.
That leap to the world beyond the lab will be the perfect prompt for you to think about submitting to the next Special Issue: Visual Search in Real-World & Applied Contexts. This will be in Cognitive Research: Principles and Implications (CRPI). The deadline is soon, but if you might have a paper for us, send me an email and we can discuss the submission date.
