Embedded Files
We explore Nelson Cowan's Embedded Processes model, which offers an alternative to the classical "stores" model of memory, and suggests a different way to interpret its function, specifically Working Memory.
While less discussed in educational circles, the growing influence of learning science on the profession suggests we're now ready to explore this model and its potential benefits for understanding memory dynamics.
Recommended prior knowledge: for context, see this previous thread Learning and memory as a cycle and the network model which I will use to describe the model below (though it is not part of the original model)
Recommended reading on the model: Cowan et al., (2024) The relation between attention and memory (open access), and see below
How do you imagine Working Memory? Take a minute to think:
Is it a temporary storage place?
Or is it a function that integrates several processes?
If you had to choose a better fit for your internal representation, which would it be?
Thinking about our memory using a model of a dynamic neuronal network, shifting between encoding, consolidation, storage and reactivation states (as described here) - How does working memory (WM) fit? Where do our understandings of WM and long-term memory intersect?
Let's explore by examining two views of WM: as a dedicated "storage", versus a dynamic "function."
This simple model highlights three cognitive functions as three “Stores”: attention, Working memory, and long-term memory (LTM) operating like a “conveyor belt”: selecting, processing and then storing information.
This model builds on the classical stores model of Atkinson and Shiffrin (1968) and the later crucial development of the model by Baddeley (2000), who added several essential elements to the WM model, but still sees the systems as differentiated stores.
*The model also focuses on attention rather than on the sensory store, to highlight the cognitive function involved.
The model can explain “cognitive load”, when input exceeds WM capacity. But it doesn't address the Attention-WM or the WM-LTM interactions: How do “items” move from temporary storage to LTM representation?
The Embedded Processes model (Cowan, 2019, 2024) offers a more dynamic and integrated view of WM: not a separate store, but a function arising from attention and LTM interaction. Let's explore it, using the network model (which is my interpretation, not part of the original EP model, see below).
Cowan Identifies an activated subset of LTM (aLTM), a small part of which is in the focus of attention (FoA).
Information from the environment is selected for and may be binded, and combined with activated representations while in the FoA.
Memory is working without a Working Memory!
See the original below and in Cowan et al., 2024 Fig. 1 (open access)
As attention fades or shifts, the new construct remains briefly in the activated (but unattended) LTM, before becoming inactive. Manipulation and the time in aLTM influence later consolidation and storage chances.
“WM limitations” stem from the limited attention capacity (3-5 items), and from limited time of LTM activation (<1min).
Information can enter aLTM without attention (e.g. unconscious priming effects), or consciously either top-down (executive function) or bottom-up (prediction error) processes.
The embedded processes model requires more knowledge and integration to grasp. Yet it is parsimonious and dynamic: we can follow one thread, from initial representation through binding and deactivation.
As for using it for educational purposes, it centers around the interaction among guiding attention, activating prior knowledge (in aLTM) and the binding process (in Foa), which are the essential elements of learning (rather than focusing on the cognitive load experience).
This model aligns well with the dynamic view of learning and memory, emphasizing activation, consolidation and reactivation. Working Memory is not a place but the critical ability to manipulate active representations, shaping memory while learning.
The original Embedded processes model from:
Cowan, Nelson, et al. "The relation between attention and memory." Annual Review of Psychology 75.1 (2024): 183-214.
Image description: Schematic representation of attention and long-term memory (LTM) in an embedded processes view. Inputs from the environment pass into an activated subset of LTM (aLTM), represented by the large, irregular shape. Some subset of this information passes into the focus of attention (FoA), which is severely limited in capacity. Solid arrows from the environment represent information entering the FoA, represented as two shapes. Knowledge from stored LTM can be used to create structures (e.g., new chunks) from stimuli currently in the FoA, enabling the information to be offloaded out of the FoA into aLTM (cloud with conjoined shapes) and stored as a new LTM. Primes presented either without conscious, explicit awareness (dashed arrow, representing input from the environment) or with awareness can activate stored concepts from LTM, which in turn can more easily pass related content to the FoA.
Read more about the embedded processes model (and how it compares to the stores model and other models too):
Cowan, N., Bao, C., Bishop-Chrzanowski, B. M., Costa, A. N., Greene, N. R., Guitard, D., ... & Ünal, Z. E. (2024) The relation between attention and memory. Annual Review of Psychology, 75(1), 183-214. Pages 186-188, including Figure 1 are focused on the model
Cowan, N. (2019) Short-term memory based on activated long-term memory: A review in response to Norris (2017). - a response to the "Stores" view:
Norris, D. (2017) Short-term memory and long-term memory are still different. Psychological bulletin, 143(9), 992. - In support of the "Stores" view
Cowan, N. (2017) The many faces of working memory and short-term storage. Psychonomic bulletin & review, 24, 1158-1170. - to realize the WM is not so simple
#ThrEduBlog Published: January 2025
Google Sites
Report abuse