“10 for $10”

Price labels like this have become a sign of the Great Recession. You're invited to buy ten of something for $10. Small print says "$1.00 each." Huh? That's right, there's no discount for buying ten items. So what's the point?
Rest assured that supermarkets must have reason to believe that these labels boost sales. The question is, how? I'm not aware of any academic papers directly addressing 10/$10 pricing. Two ideas come to mind.
One is that it's a twist on the old "economy size" tactic. Many shoppers instinctively reach for the bigger package — or shop at the big box store — in the belief that they're getting a better deal. Sometimes they are, and sometimes they aren't. In this case, shoppers might lob ten cans into their cart and never read the fine print.
I doubt too many people do that, though. It seems that 10-for-$10 pricing is mostly applied when you wouldn't normally buy ten of an item. I don't think I've ever bought ten cans of anything, save soft drinks. My impulse would be to scrutinize the label to see whether you can buy less than ten at the special price — which you can.
An alternate explanation for 10/$10 prices is anchoring, the cognitive phenomenon described by Amos Tversky and Daniel Kahneman in the 1970s. Anchoring occurs when we guesstimate numerical quantities. Shoppers do that a lot. The shopper must decide, how much should I pay? and how many should I buy? Strict logic cannot answer such questions. The answer is necessarily intuitive (and often, must be made while wrestling kids or yakking on the cellphone). It's been shown that when people have to come up with a numerical estimate on intuitive grounds, they are easily swayed by any numbers provided as cues (or "anchors.") With 10/$10 pricing, the suggested ten items becomes an anchor. It sends the subtle message, of course you need ten cans of salsa this week! On the face of it, that sounds ridiculous. But anchoring exerts an effect even when we know the cue is absurd.
One of the most amusing anchoring experiments, by Karen Jacowitz and Daniel Kahneman, asked a group of subjects this two-part question:

(a) Does the average American eat more or less than 50 pounds of meat a year?
(b) How much meat does the average American eat in a year?

The group's median answer was 100 pounds of meat. The researchers asked a separate group a similar question, except that this time, part (a) asked whether the average American ate more or less than 1000 pounds of meat. For this group, the median estimate was 500 pounds. Just hearing a different cue — 1000 rather than 50 pounds — boosted the group's estimate five-fold. Furthermore, anyone who thought about it could have calculated that 1000 pounds was a ridiculous answer. It would be the equivalent of ten McDonald's Quarter-Pounders a day! The ridiculous anchor value still had a huge statistical effect on answers.
In the experiment, the subjects were simply guessing the answer to a trivia question. Shoppers guess too in estimating how many cans of salsa their family will eat. It's possible that 10-for-$10 pricing has an effect similar to the cues in Jacowitz and Kahneman's experiment. It makes shoppers buy more items than they would have.

SEE ALSO
Jacowitz, Karen E., and Daniel Kahneman (1995). “Measuring of anchoring in estimation tasks.” Personality and Social Psychology Bulletin 21, 1161-1166.