2.3 Introduction to Memory

Memory isn't one single thing. Psychologists sort memories by whether you can put them into words and by what they're about.

• Explicit (declarative) memory: memories you can consciously recall and put into words.
• Episodic memory: an explicit memory of a personal event, tied to a time and place, like your last birthday party.
• Semantic memory: an explicit memory of general facts and knowledge, like the capital of France or the meaning of 'neuron.'
• Implicit (nondeclarative) memory: memory that shows up in behavior without conscious recall, and is hard to put into words.
• Procedural memory: an implicit memory for skills and procedures, like tying your shoes or typing.
• Prospective memory: remembering to do something in the future, like taking medicine at 8 p.m. or texting a friend after class.

The multi-store model (also called the Atkinson-Shiffrin model) says information passes through three stores. Sensory memory holds a brief, detailed copy of what your senses take in: iconic memory for sights, which lasts a fraction of a second, and echoic memory for sounds, which lasts a few seconds. Information you pay attention to moves into short-term memory, which holds a small amount for less than a minute without rehearsal. Information that is rehearsed or processed further is encoded into long-term memory, which has a huge capacity and can last a lifetime.

The model also separates two kinds of encoding. Automatic processing happens without effort, like remembering what you ate for lunch or where on a page you read something. Effortful processing takes attention and work, like studying vocabulary.

The working memory model sees short-term memory as an active workspace, not just a waiting room. Working memory has several parts. The central executive directs attention and coordinates the other parts. The phonological loop holds and rehearses sounds and words, like repeating a phone number in your head. The visuospatial sketchpad holds images and spatial layouts, like picturing the route to a friend's house.

Because the phonological loop and the sketchpad are separate, you can do one verbal task and one visual task at once more easily than two verbal tasks. That's why listening to a podcast while reading a book is much harder than listening while you draw.

The levels of processing model says how deeply you process information determines how well you remember it. There are three levels, from shallowest to deepest. Structural processing focuses on appearance ('Is the word in capital letters?'). Phonemic processing focuses on sound ('Does it rhyme with weight?'). Semantic processing focuses on meaning ('Does it fit the sentence: She ate a ___?'). Semantic processing produces the best memory, which is why connecting ideas to their meaning beats rereading.

Long-term potentiation (LTP) is the strengthening of a synapse after repeated activation. When one neuron repeatedly helps fire another, the connection between them becomes more efficient, so the receiving neuron responds more strongly next time. This physical change in synapses is thought to be a basis for learning and memory. The excitatory neurotransmitter glutamate plays a key role in LTP.

Different brain systems also support different types of memory. The hippocampus is key for forming new explicit memories, while the cerebellum supports some procedural learning. Famous case studies of people with hippocampal damage show this split: such a person can steadily improve at a new motor skill across days of practice while having no memory of ever practicing it.