For the committed memorizer, my last blog post was a downer. Perhaps you suspect I was not giving you a completely accurate view of the real capacities of human memory. In the spirit of candor, I present the USA Memory Championship. A journalist, Joshua Foer, who attended the yearly event reported:

“[Contestants] were memorizing hundreds of random numbers, looking at them just once. They were memorizing the names of dozens and dozens and dozens of strangers. They were memorizing entire poems in just a few minutes. They were competing to see who could memorize the order of a shuffled pack of playing cards the fastest.”

Foer interviewed a contestant from England who demurred, “I'm not a savant. In fact, I have just an average memory. Everybody who competes in this contest will tell you that they have just an average memory. We've all trained ourselves to perform these utterly miraculous feats of memory using a set of ancient techniques.”

“We've all trained ourselves to perform these utterly miraculous feats of memory using a set of ancient techniques.”

 “Whoa! I thought so,” you say. “Tell me about those ancient techniques!”

 I promise I will, but one more example and a bit about the brain first.

 In 1909, as part of a chess exhibition, the Cuban chess master José Raúl Capablanca played 28 chess games simultaneously. Capablanca rotated from one board to the next taking two or three seconds to make his move at each board. He won all 28 games.

Capablanca was called the “Human Chess Machine.” Many assumed he had a photographic memory of each board that he accessed as he moved from game to game. Capablanca, however, claimed “I see only one move ahead, but it is always the correct one.”

Chess is unique as a tool in studying human cognition. Chess players are given a numerical rating (ELO rating) based on the ratings of others they have played and who won the game. The ratings generally predict with significant accuracy who will win a given match.

In one kind of cognitive study chess players were shown a chess board briefly (3-10 seconds) and then asked to recall the position of the various pieces. In the graph to the right the dashed line shows that there is a linear relationship between skill in chess and the number of pieces the individual could recall. This was only true if the position of the pieces reflected actual games. If the pieces were randomly placed, skill gave only a very slight advantage.

The researchers who conducted this study argued that the data “offers strong support for chunk-based theories [of memory].”

Let me briefly explain what a chunk-based theory of memory means. Working memory has a very small limit in the number of items it can hold simultaneously. The number is currently believed to be 4 ± 1. With that reality, how could chess grand masters recall the positions of nearly 20 chess pieces with only a 30 second rehearsal? The answer is that they didn’t retrieve those 20 pieces individually; they retrieved that cluster of 20 pieces using logic.

Philip Ross in Scientific American puts it this way:

“To a beginner, a position with 20 chessmen on the board may contain far more than 20 chunks of information. . .A grandmaster, however, may see one part of the position as ‘fianchettoed bishop in the castled kingside,’ together with a ‘blockaded king’s-Indian-style pawn chain.’”

This compresses the position of the pieces on the board into a much smaller number of chunks that will fit within the limits of working memory. Perhaps three chunks:

1. fianchettoed bishop

2. castled kingside

3. blockaded king’s-Indian-style pawn chain

Chunks are a form of data compression—a zip file. A chunk is a group of items that are treated as a single item under a specific label in long-term memory. When it is called up to working memory to solve a problem, or for possible modification, it counts as only a single item.

A chunk is a group of items that are treated as a single item under a specific label in long-term memory.

Working memory creates chunks and uses existing chunks that it recalls from long-term memory. Chunks can be unpacked if needed, but to do so while in working memory risks cognitive overload. You might picture your packed suitcase as a chunk and the mess that results if the TSA decides to search it. You need a worktable large enough for the items that are no longer packed (or the floor)!

At the most fundamental level, chunks consist of concepts, not just items. The difficulty most people have in remembering names serves as a powerful illustration of the profound difference.

We have trouble remembering names because they are empty labels—not concepts. Because the brain stores concepts and not hollow labels, it frequently betrays us when we try to remember the name of someone we met in passing. This has been called the Baker-baker paradox.

The Baker-baker paradox shows us we remember concepts—not labels

Baker as a surname is an arbitrary label, while baker as an occupation is a rich concept. We can easily visualize a baker in action at a bakery crafting scrumptious baked goods from flour, butter, etc. using an industrial mixer. We can see her or him kneading dough, and we can delight in the smell of the baking confections. It’s not hard to remember all that goodness, but seldom does someone strike us as looking like a Baker.

Chunks consist of concepts, but chunks are not all the same size. When we enlarge a concept through collecting additional examples that add nuance and depth, we are chunking. Chunking also occurs when we logically link an existing concept with other concepts. Perhaps these are new correlations, or perhaps we now recognize something about causality which provides crucial insight. Chunking can also create complicated webs of logically related concepts that form schemata, or models, or systems.

Here’s how neuroscientist Daniel Bor describes the power of chunking:

“Closely connected individual. . . [concepts] form chunks together, which then connect up themselves in ever larger bound objects in memory. In this way, we. . . create a highly functional, hierarchical, interrelated bank of knowledge, where, by the time we reach adulthood, most seemingly novel items have some preexisting context. And these heavily embedded prior expectations from the fruits of our vast learning can in turn heavily guide our attention to decide what to load into our working memory, furthering our chances to discover in awareness something novel or important by which to incrementally improve our world model.” (emphasis mine)

We all chunk, but we don’t necessarily do it well. Many chunks are crude hacks. Here are a few from Wikipedia.

• Parts of the brain associated with memory, Herds of Animals Cause Panic. Hippocampus, Amygdala, Cerebellum, & Prefrontal Cortex.

• 3 types of brain encoding: SAVE (Semantic encoding, Acoustic encoding, Visual encoding.

• ·Layers of the Open System Interconnection Model of computing or telecommunications:  Please Do Not Teach Students Pointless Acronyms – with each of the initial letters matching the name of the OSI layers from bottom to top (physical, data link, network, transport, session, presentation, application).

Here’s one from the UNC Chapel Hill Learning Center website:

Order of operations in mathematics: Please Excuse My Dear Aunt Sally (parentheses, exponents, multiply, divide, add, subtract).

These are all examples of what I call clunky chunks. It is most likely that what will endure in the memory is the mnemonic device without the items it was constructed to scaffold.

Where there is a logic that binds the items in the chunk together, we must seek for that rationale.

We must insist on understanding wherever possible and not be content with mere association. The common distortion that we memorize first and perhaps understand later betrays a profound misunderstanding of how the brain works. Memory is a byproduct of understanding.

I promised you the “ancient techniques” for “miraculous feats of memory.”

Here’s the short version (with a little context) from journalist Joshua Foer who ended up winning the USA Memory Championship!

“We often talk about people with great memories as though it were some sort of an innate gift, but that is not the case. Great memories are learned. At the most basic level, we remember when we pay attention. We remember when we are deeply engaged. We remember when we are able to take a piece of information and experience, and figure out why it is meaningful to us, why it is significant, why it's colorful, when we're able to transform it in some way that makes sense in the light of all of the other things floating around in our minds. . .”

“. . . these memory techniques -- they're just shortcuts. In fact, they're not even really shortcuts. They work because they make you work. They force a kind of depth of processing, a kind of mindfulness, that most of us don't normally walk around exercising. But there actually are no shortcuts.” (emphasis mine)

It takes work to remember for the sake of impressing others with your memory, or winning memory contests. If you want to remember the first 100 digits of Pi, you will have to create a chunk to scaffold the order of the digits. There are no cheap tricks, only clunky chunks!

Your mind is too astounding to devote to spinning its wheels in the mud. Harness your latent creative chunking potential to solve problems that matter, and memory will follow naturally. Remember that the supreme purpose of memory is “using the past to intelligently guide decision-making.”

Joshua Foer and the USA Memory Championship. The summary quotes above start at 17:56. The YouTube video contains a couple of objectionable pictures.

José Raúl Capablanca

Chunking mechanisms in human learning, Fernand Gobet, Peter C.R. Lane, Steve Croker, Peter C-H. Cheng, Gary Jones, Iain Oliver and Julian M. Pine, TRENDS in Cognitive Sciences, Vol.5, No.6, June, 2001, pp. 236-243 https://doi.org/10.1016/S1364-6613(00)01662-4

Philip Ross, The Expert Mind, Scientific American, August, 2006, p. 69.

Baker-baker Paradox: William Scott Terry, “On the Relative Difficulty in Recalling Names and Occupations,” The American Journal of Psychology, Spring, 1994, Vol. 107, No. 1, pp. 85-94.

Daniel Bor, The Ravenous Brain, 2012.