Cognitive Daily

Does this ever happen to you? You’re preparing green beans to be cooked, putting the stems in the trash and the beans in a bowl. Suddenly you realize you’ve started putting the stems in the bowl. The dinner guests will be arriving soon, and now you have to search through the beans to pull out the stems, in order to avoid an embarrassing incident later that evening.

Okay, maybe it’s just me. But what’s the best way to find the stems? Is it faster to pore over the bowl, methodically scanning for each remnant? Or is it better to step back and take a holistic view of the bowl, letting the stems “pop” out of a sea of green?

People who study visual search have found anecdotally that just “relaxing” and looking for objects based on “gut instinct” can often be more effective than actively directing attention to a search. Jeremy Wolfe calls this “relax” strategy “using the force.” You can try it out. You’ll be looking at a figure with two types of shapes:

Most shapes look like the one on the left: a circle with two gaps. Your job is to look for a circle with just one gap, and say whether the gap is on the left or the right. Now try it using the relax strategy on the image below:

Next try the same task, but this time use the active search strategy: actively and systematically search for the object. Remember, your job is to say whether the circle with one gap is facing left or right.

Which was easier? Let’s make this a poll:

For years, there has been a general sense among researchers that relaxing can make visual searches more efficient, but no scientific study had been done until recently. A team lead by Daniel Smilek asked volunteers to search figures like the ones above. They systematically varied both the number of objects in the search and difficulty of the search (by increasing the size of the gap and making it easier to spot). Half the volunteers were instructed to relax and just let the object pop into view, and the other half was told to search actively and methodically. Here are the results:

For easy searches, there was no difference in search strategy. But when the searches were difficult, the relax strategy was significantly more efficient: reaction times were faster. More importantly from a visual search researcher’s perspective, the slope of the reaction time line was smaller for the relax strategy, which means the search is more efficient — it’s more effective as the task gets more complex.

So relaxing, or “using the force,” is indeed more effective for this type of search. Will it work better in every situation — everything from looking for lost keys to scanning a document for typos? Smilek’s team says probably not: a variety of strategies can be effective, depending on the precise circumstances of the search. But for the first time, they document a case where just relaxing appears to be more efficient than active searching. They argue that the reason the relax search works is because people are relying on executive control: the automatic processes in the brain for handling spatial computations, rather than an active, cognitive search.

Smilek, D., Enns, J.T., Eastwood, J.D., & Merikle, P.M. (2006). Relax! Cognitive strategy influences visual search. Visual Cognition, 14, 543-564.

Comments

#1 Alex Whiteside
October 31, 2007

I found the active search far quicker, but that’s mostly because I was doing left-to-right, top-to-bottom searching (like reading) and it’s the second circle along that way!
#2 David Drake
October 31, 2007

I’m curious as to why a relaxed strategy should be more effective, if that is the case. What provides the efficiency? I suppose it only emphasizes how much is happening outside of the conscious stream of thought while still profoundly influencing it.
#3 Aaron Lemur Mintz
October 31, 2007

I had the same result as Alex. My anecdotal hunch is that this kind of habit happens for a significant number of people, and so you can estimate how long it’ll take people to find object based on where it is spatially.
#4 speedwell
October 31, 2007

I’m old enough to have held that archaic, obsolescent position known as “proofreader.” You can’t do the job without this.
#5 marciepooh
October 31, 2007

Proofreaders may be obsolete but, unfortunately, speedwell, the need still exists. Spellcheck is not the same. Just read my local paper.

I screwed up and did the same thing both times. I think my method is somewhere in between. I scan the whole picture but don’t look at each circle careful. There is no particular pattern to the scanning.
#6 Theo Geer
October 31, 2007

I’m all about the relaxing tactic. — If you want a larger personal test to see what works for you, do a pair of word-search puzzles. Time yourself.

In the first one, look at the first three words and scan left to right row by row, examine the surrounding characters each time you find the first letter of the word(s) you’re looking for. Proceed until you have found each of the words.

In the second one, start with the first three words and relax your gaze. just let your mind try to find the words you’re looking for.

In both, if you feel you’ve spent too much time on a word or set of words then move on and come back to that one later.

I find the ‘relax’ method often goes much more quickly than the comprehensive search method. – I often combine them both for even better results. (Relaxed search until a word is obviously not popping out, then I do a comprehensive search).
#7 Curt Rozeboom
October 31, 2007

I’ve been using this technique for years on word-searches, myself, and never understood how I was actually doing it. I thought it was just a matter of practice, but it turns out I merely stumbled upon an innate mental capability. Cool.
#8 Brian Friel
October 31, 2007

You’ll probably find that the ‘relax’ method will work best for most people (it did for me, at least). It seems to me that the ‘relax’ method is just another name for employing parallel visual search processes, whereas other ‘search’ strategies tend to involve more serial processing. For the ‘relax’ strategy condition, the shallower RT slope over levels of task difficulty seems to suggest that this is the case.
#9 marciepooh
October 31, 2007

Theo, that is exactly how I do word-searches.
#10 Brian
October 31, 2007

Keep in mind that the difference between the relax and active search times weighs in at around 200ms, which is huge by researcher standpoints, and “just a fifth of a second” for the subjective experience standpoint.

As far as mechanisms are concerned, a potential source of this effect is what is known as ‘saccadic suppression.’ This is a pretty well-researched concept, and essentially describes the finding that when you make a quick eye movement, visual information is ignored while your eyes are in motion. The more eye movements you make, the longer you spend ‘ignoring’ the search array. A recent paper from McCarley et al. demonstrated a similar finding: participants detected changes in moving stimuli more frequently when they moved their eyes less frequently.

Saccadic suppression is, I’d say, a pretty likely candidate for this effect. There is also quite a bit of evidence that planning eye-movements can be cognitively demanding, so it’d be unsurprising if this also contributed to Smilek and friends’ results (i.e. the ‘executive control’ explanation they propose).
#11 Dave
October 31, 2007

“Relax” method may work better initially – ie you may find the first few things quicker – but I wouldn’t trust it to find everything – for that you need to do a systematic search. ie the bean stem example – you better be doing a systematic search for that one.

I guess I was confused by the test – you should have emphasized that there was one item to be found. Because using “relax” method, after finding the item, I keep jumping around and looking for others – significantly slowing me down.

So I guess a qualifier is needed – for finding a known number items a “relax” search would be better. To find an unknown number you have to get systematic…
#12 raj
October 31, 2007

Yep, I agree with Alex. The active search part of the test would have been far more representative if the object for which I were searching didn’t just happen to be near the upper-left corner. Since most people reading this web page, read a language which progresses left to right, top to bottom, I would bet that’s the way most people would search also.
#13 Martin La Belle
October 31, 2007

I agree with most of the above. The premise is sound but the experiment was not very effective. Perhaps an experiment run in a java, or Flash environment would be more effective, while at the same time raising the interest in such phenomenon.
#14 Moopheus
October 31, 2007

Proofreaders are not obsolete. I still do it. In fact, automated spellcheckers and the like practically guarantee my full employment for the foreseeable future. But indeed, once I have finished my first, perfectly serial pass through a set of proofs, I page through a second time and let my eyes scan quickly and semirandomly over the pages again, and invariably I find a few typos I missed the first time.
#15 Martin
October 31, 2007

Signal detection theory predicts that motivation of hit or miss, ie. the frame of the search, will influence the results. Who knows how motivation affects the searchtypes…
#16 P.Tane
October 31, 2007

I found the active search to be faster and easier. I accidentally voted for “no difference”.
#17 BDuse
October 31, 2007

You just improved my Bejeweled game a ton, thanks!
#18 Warren
October 31, 2007

But what’s the best way to find the stems?

Pour cold water into the bowl. The stems will float to the top.

(Yes, I read the post; I’m just being a smart-ass.)
#19 Nigel Thomas
October 31, 2007

I voted for active search, because it went quicker for me, but, like many other commenters here, I think that was probably just because the target happened to be near the upper left and it is natural for readers of the Roman alphabet to start scanning from there. What is remarkable is that the “just relax” technique works so well. I certainly still found the target quite quickly that way.

I wonder, however, if, at least for targets and distractors like these, where the difference is quite subtle, if the two techniques really differ that much. I certainly did not get the sort of instant “pop-out” that one experiences with truly distinctive targets (like, for instance, a green shape amongst red ones). Perhaps I was doing it wrong, but it seems to me that even when I was trying to “just relax” I was still examining individual circles. It differed from “active search” in two ways, I think: on the one hand I was choosing circles to examine at random rather than following a systematic order, and on the other hand I was not scrutinizing the circles as carefully as I did when I “actively searched,” but just sort of letting my eye drift quickly over them. The first of these differences would seem likely to make “just relaxing” less efficient, at least in circumstances like this, where there are a lot more distractors than targets. Of course one might just happen to get lucky, but more likely one would waste a lot of time by revisiting the same distractors repeatedly. On the other hand, the second difference would tend to speed thing up (and perhaps when there are not too many more distractors, this effect predominates). Perhaps the most effective strategy would be still to scan the attention over the area in a systematic way, but not to bother looking too long and carefully at the individual items (as I, at least, tended to do when “actively searching,” making sure, as it were, that each one was a distractor and not the target) but have faith that the target will indeed “pop-out” once you are looking at it.
#20 Stephen Downes
October 31, 2007

Use the force? Now *there’s* some good peer reviewed advice.

Did the authors actually say ‘use the force’ or are you just embellishing? Oh, there’s no way to tell – there’s no link top the article. It must be locked behind a paywall.

Peer reviewed. Pah! If it isn’t open, it wasn’t reviewed.
#21 Kimmitt
October 31, 2007

One of my hobbies is playing strategy games with pieces which look very similar but have nontrivially different effects and often need to be placed in specific parts of the board (to simulate the historical events in question).

I definitely use a combination of strategies — the “relax” strategy for sorting out categories of pieces (pieces marked for being placed in a particular section of the board) and direct scanning for finding specific pieces.
#22 Jp
October 31, 2007

Been doing this with crosswords for ages. Shame I could never figure out a way to move my furniture with it though.
#23 Dave X
October 31, 2007

I use the relax method for finding four leaf clovers, but my goal in doing the search is usually to relax so a systematic search wouldn’t serve. Being able to hand someone a four leaf clover is a just a bonus.
#24 Tim
November 1, 2007

Fascinating topic. The faster the better. For me, I first use the relax method since it will be faster if visual object is easy for me to recognize. This was the case with widely spaced symbols provided in example. If I whiff with the relax method or judge the visual search to be too difficult, I switch to the systematic search.

This topic seems to speak to visual processing speed. There are likely many variables involved. What is the correlation between visual recognition speed and objective measures of “success” in life?
#25 tom B
November 1, 2007

One imagines this all has to do with the avoidance of being eaten in a complex setting, like a forest.

Perhaps jigsaw puzzles could provide the basis of some interesting studies.

Interesting aside: my wife has an amazing talent for picking out 4-leaf clovers.
#26 david1947
November 1, 2007

Not enough samples. The “force” one was way down near the bottom, the linear “active” scan one up near the top. I’m not sure the “force:” is not just a defocussed scan, and still therefore at least a top-down process. One needs also to account for the practice effect. How often do we need to search for singly broken rings in a field of doubly broken rings? So the samples should also use different images to scatter that effect.

I happen to agree with the hypothesis – grokking is faster – but this example clearly used a pathological data set that gave counter results.
#27 Dave Munger
November 1, 2007

Okay, after the first round, the Active strategy had a slight lead. But of course, some people claim the Active graphic was easier. So just now I switched the graphics and posted a new poll.

Now the graphic that *was* used for the Active search is used for the “Relax” search. We’ll see if the results are different.

(Here are the results from the first poll):

Relax (154)
Active search (169)
No difference (128)
#28 Freiddie
November 1, 2007

I’d say it’s better to just use random images, so that it’s unbiased. Perhaps you could display a different image (from a random collection of images) for each person visiting the website…
I personally don’t really know the difference between the two types of searching in this scenario because the search task was a bit too easy. In harder (and everyday) scenario, I would be searching by instinct first, then use the active searching method just to finished off the search (in case I didn’t find what I was looking for or when I’m trying to get rid of ALL the stems from the bowl).
#29 Kevin
November 1, 2007

It seems to me that anxiety should enter into the equation. If there was a time limit in which to find the 1 gapped circle versus no time limit, my guess is that finding the circle with no time limit would be easier.

If there is some sort of order to the ‘Active’ search there should be no difference between ‘Active’ and ‘Relaxed’. However, when anxiety is introduced the ‘Active’ search would probably become less orderly and more chaotic and as a result would take much longer than ‘Relaxed’.
#30 Aaron
November 1, 2007

I used relax on both and the first was easier. Why? Because I am trained to start reading from left to right and top to bottom, so naturally I caught the first one very quickly. I don’t think this test is well designed.
#31 Dave Munger
November 1, 2007

That’s a good point, Aaron — which is why we switched the two images a couple hours back. If you look back at comment #27, you’ll see that the Active strategy was slightly favored before. Now, with the pictures switched, Relax is trouncing Active.

Freiddie — in the actual study, the images were indeed randomized. Our little demo is just designed to give you a rough sense of how it works.
#32 Freiddie
November 1, 2007

OK.
#33 ValiumFace
November 3, 2007

No significant difference, although the I do use the ‘relax’ method for other things and it does seem quicker. I think just relaxing and being in the present moment is the fastest way. Stress or pressure of any sort causes me to become slower, even if I think I’m being more active.
#34 al1027
November 3, 2007

Does *anybody* remember the Yerkes-Dodson law??? Anybody? The authors didn’t mention it, nobody in this blog has mentioned it… Aside from all the baggage that comes with using a visual search task, this study is essentially confirming an observation that was originally reported in 1908. 1908!!! While “using the force” may be appropriate for the behavior, “reinventing the wheel” seems to fit for the study as a whole.
#35 Silicon.shaman
November 5, 2007

I generally find that “using the force” works in a wider context as well. Relaxing and taking a holistic approach works when trying to find *anything*, even if one is searching one’s memory for something.
#36 Phil
November 10, 2007

For me, using the relaxed approach is almost always a more effective strategy. But, then, I’m an asst. professor in art. So, looking is something I’ve learned to do.

I think that one of the most difficult things to teach beginning art students is how to look passively while remaining conscious of how that looking is taking place. It’s not (I think) that it’s more difficult, it’s just a new concept for them.

I’d like to suggest that respondents’ visual experience will have the strongest influence on the results of the poll.
#37 Karen
November 18, 2007

What about scanning the field for multiple objects of unknown number? One would seem to have to use the active method to be sure that all of the sought object have been found.