Schematic Organization

Schemas are based on what similar situations have in common; because no two situations are ever exactly alike, schemas Opens in new window must be somewhat flexible: their elements, that is, the categories of objects, single events, actions within a scene or event, are variable within certain limits.

This flexibility makes a schema somewhat like an equation, with abstract variables that can be filled in with different values. These variables, in turn, are like “slots” into which various particular features can be fitted, provided they do not violate the basic nature of the schema.

This flexibility is one of the things that make schemas so useful, because they can be adapted to many different situations.

Note that in order to correctly deal with the restaurant schema, not only do we have to be familiar with the temporal organization of the schema itself, but we have to recognize instances of the categories of table, chair, waiter/waitress, cash register, and so on, and that we must know what to do in response to these instances.

All of these have a role in the schema, and all of these can differ considerably in detail, and still evoke the restaurant schema.

The organization of musical time into repeating cycles of meter works in a schematic way.

The downbeat is a very important type of category in metrical structure. It is a beat that marks the beginning of a metrical cycle, and functions to maintain the cyclical grouping standard of meter.

The downbeat is like a schema slot, in that many different kinds of musical events may occupy the downbeat position in the measure and still function as downbeats.

The category of downbeat actually has a graded structure, with some types of events being more prototypical or better examples of the downbeat than others.

The categories that comprise a schema may also have “default values” (Barsalou, 1992). Details of a situation that are not noticed as unusual quickly drop out of awareness and their slots are assigned default values in memory.

Default values fill in typical or average instances in memory for variables not attended to at a particular time—creating a kind of generic reality.

In our memories of things that have taken place outside, for instance, the sky is usually there, although we may not have looked at it at all. The sky slot may therefore have a default value.

Based on the category prototype effects mentioned earlier, schema default values are probably responsible for many episodic memory Opens in new window distortions.

When recalling a particular episode, we often do not remember which parts of the experience we were actually paying attention to, therefore we do not know which parts of the memory Opens in new window may be our own default values and which parts “really” happened.

We have filled in the unattended-to parts of the memory with our own previously established prototypical values. In this way, different generations of memories can get conflated.

In the very long term, this probably causes the details of episodic memories Opens in new window to slowly drift in the direction of default values (J. Mandler, 1984): our memories slowly modify themselves to become more and more like the most prototypical instances of our own categories and schemas.

Because our categories Opens in new window and schemas Opens in new window themselves are evolving or being updated based on new experiences and expectation failures, this means that subsequent experience can actually modify our older memories.

Schemas and Normalcy

We are most likely to notice and remember things when they do not fit exactly within our schemas, and these novel episodic memories will tend to be connected by association with a relevant schema.

This means that noticing and remembering are most likely to take place when our expectations fail.

We would tend to remember these types of events, and would also tend to be reminded of these episodes when something in a later situation evoked the same schema.

From the perspective of biological adaptation, when our expectations fail is when we most need a specific (episodic) memory of another failure of the same expectation to help us figure out what to do.

That is, we noticed and remember things when expectations fail because these types of situations could eventually require that a schema be modified, or in more extreme circumstances, that a new schema be formed.

If a particular set of expectations fails a number of times, the schema for that particular situation will need to be updated.

Hence specific episodic memories Opens in new window representing the failure of particular expectations (schemas) are connected to those expectations by association: they are available to remind us at the relevant moment when we might need to consider updating a schema (Schank, 1982).

All of our schemas taken together form a kind of averaged model of how the world usually is, a kind of generic reality. Without this complex of frameworks, the world would be always strange and unfamiliar, and we would spend all of our time trying to figure out where we were and what was going on.

Most of the time, this model works so well that we do not need to pay much attention to many details of our lives. When something deviates sufficiently from the normal, it receives our attention; it is remembered and this memory is associated with the relevant schema in memory.

Because they are part of semantic memory, different levels of schemas may be hierarchically arranged, moving from schemas about generalized aspects of reality to very specific schemas about particular kinds of objects or events (remember, however, that the internal organization of a particular schema is usually based on the organization of the scene or event it represents).

The actual restaurant schema itself, however, would derive its time sequence from the sequence in which the events typically happen.

In relation to a particular piece of music, we might have more general schemas about the large-scale organization of a piece or its musical style, and we might have more detailed schemas about the local arrangement of sections, particular types of pattern in those sections, and so forth.

The disadvantage of a generalized, schema-driven recognition system is that over time, because they are always undergoing the process of being slowly generalized, particular episodic memories of experiences that are similar and fit the same schemas tend to become confused.

That is, when we have any type of experience repeatedly, we have great difficulty remembering the details of any particular occurrence, unless they are fairly unusual. This is referred to as an “interference effect” and is based on the idea that similar memories interfere with each other.

Interference effects are a direct result of the limits of categorization: we simply cannot retain all of the minute differences between different but very similar experiences. Music that uses repeated, highly similar patterns is designed to produce this effect.

Because a schema has been defined here as a higher-level categorization of experience, and because categories create boundaries, an interesting question arises.

What constitutes the boundearies of a temporal schema?

One answer to this question that has been proposed is that schemas are bounded by the achievement of goals (Schank, 1982).

This is a kind of high-level grouping of real-life experience. Certainly, in real life, these schemas and goals are nested, and there can be a number of both short- and long-term goals in operation at the same time.

In linear, goal-directed music, which is far less complex than real life, there may still be a number of hierarchical levels of goals operating at a particular time. It is often the case in such music that the arrival at a goal of some sort signals the end of a section—in essence, a temporal category boundary.