June 14, 2015

Remembering Life Stories (2): Temporal Content

mnemonic reconstruction of storylines depends on temporal information in trace memories

The previous post outlined three components of a complex dynamic that facilitates efficient remembering of lengthy and elaborate biographical storylines. Today's post aims to unpack what I said of the first component:

Some content self-sequences -- Cognitive memory research shows that information itself can dictate the mnemonic construction of relative times. In focusing on Character, we find temporal structure in self-sequencing memories of human growth (biological & psychological) and personal development (social, cultural, legal, political).

First we must first genuflect to the science of memory with a proper technical description of “constructive remembering”, the prevailing view of memory in cognitive studies.

Rather than retrieving memories like opening a file folder, or playing a video, we actually reassemble our memories from ‘bits and pieces of information’ (sometimes called ‘traces’ or ‘engrams’) scattered across parts of the brain (Schacter, 1996). Psychologists avoid the “storage” metaphor, but describe these trace memories as ‘preserved’ or ‘encoded’ in memory, acknowledging “the metaphor of construction implies some raw material” (Neisser, 1967). Each time these 'traces' and their associations come back to the surface of consciousness, the mind needs to (re)contextualize that information, but since the available pool of related information has inevitably changed over time - with new learning, forgetting, an overactive imagination, or just a gradual decrease in vividness - any present reconstruction amounts to an update, a recompiling if not a total rewrite of “the” memory. Specific adjustments can be disturbing and sensational, or barely perceptible, but remembering always involves both “raw material” and “construction”. First, the underlying content of memory (the engrams, the encodings, the traces, the bits and pieces, the information, or what I’ll refer to for convenience sake as the “data”) may preserve errors, distortions, and misinformation, as often as accurate data, but encoded ‘bits’ of mnemonic ‘content’ are where constructive memory always begins. From there, our minds employ various methods (frameworks or patterns or ‘schemas’ or catch as catch can) for putting content (back) together, some of which involve chronological order (Eyesenck & Keane, 2010). Now, when mnemonic rebuilding is focused on temporal sequence, these two layers of constructive remembering align naturally with “content and structure” as per the narratological concept of fabula, the version of a story that gets (chronologically) reassembled in memory. This should not be surprising because the constructivist paradigm began with a study of people remembering versions of a story (Bartlett, 1932).

Now, having affirmed textbook doctrines of cognitive psychology, let’s challenge some of its popular dogma.

Memory experts assure us that remembering is always constructive and frequently unreliable, and that unreliable memory derives from constructedness and/or from misinformation in preserved ‘memory traces’ (see above). They affirm that memories can be accurate, but remembering accurately requires both “good information” (bits and pieces of trace memory) and “good processing” (putting those pieces together correctly), whereas inaccurate remembering can result from either bad information or bad processing (improper reconstruction). By this logic, “unreliable” and “constructive” cannot be synonymous. Yes, the discussion and literature of cognitive scientists often seems to imply that memory construction is automatically unreliable, but a closer reading assures us this is not always the case (Cf. Ost & Costall, Misremembering Bartlett, 2002).

My point is not merely to insist that constructed memory is sometimes reliable but to draw out the overlooked and more intriguing point: Reliable memories involve reliable construction, just as false memories are most insidious when they connect bits of data in a trustworthy manner.

We must emphasize that “reliable construction” is not strictly a euphamism for “accurate remembering”. While the "data" of trace memories can be accurate or inaccurate, their recontextualized information (the mnemonic reconstruction) is more appropriately judged as being plausible or implausible, as infamously evidenced by those compelling false memories. We are dealing with two related dichotomies - mnemonic content and mnemonic structure - and by cross-comparison we can derive four distinct categories: (1) bad traces poorly constructed, (2) bad traces plausibly constructed, (3) good traces poorly constructed, and (4) good traces plausibly constructed. Only the fourth category renders "accurate remembering", but the fourth and the second combinations both render plausible constructions, each of them based on content believed to be accurate.

From my perspective, the second kind (bad traces plausibly constructed, a.k.a. false memories) is potentially the most helpful, and categorically the most misunderstood. While memory’s critics lump together all three “bad” types of remembering, memory’s apologists are always discussing the holistic accuracy of type four. But for us to examine reliable construction (to isolate that variable, so to speak) we cannot bother distinguishing between good and bad “data”. For the duration of this exercise, therefore, we must accept false and true remembering as if they were just alike, suspend our focus on “accuracy” and reliability, stop questioning material, and look merely at process.

In particular, because the larger project (here, now) is all about remembering timelines, we’re going to focus specifically on mnemonic reconstruction of chronological sequence. Finally, to maintain our deliberate distance from causality and “emplotment” (see previous posts), we’re going to remain focused on biographical narratives. Altogether, therefore, the central question is this: When a plausible reconstruction is assembled from whatever we think we can recall, how does the process of constructive remembering bring together trace bits of story content and build them into a temporal sequence?

Or to put that more succinctly, How does the mind reconstruct story-sequence in a lengthy, elaborate fabula? I mean, in terms of constructive remembering, actually HOW?

As it happens, one researcher has already laid down a framework for answering this question.

~~~~~~~~~~~~~~~

William Friedman is a research psychologist who stands firmly in line with the constructive viewpoint in his work on remembering “time” (which I discussed previously here). His landmark 1993 paper “Memory for the Time of Past Events”, summarized a body of prior research, critiquing eight theoretical systems of mnemonic chronology and finding that demonstrably effective methods of preserving temporal sequence (across all theories) involved reconstructive processes. In his own most succinct statement of thesis: “[O]ur chronological sense of the past is the product of an ongoing constructive process in which we draw on, interpret, and integrate information from our stored knowledge of time patterns and other general knowledge about time, the contextual associations of particular memories [and other types of associative connections]” (the underscoring is mine).

Let’s elaborate on these three aspects of remembering chronology, according to Friedman.

First, remembering chronology requires construction. While many psychologists view “constructive remembering” as a process of filling in gaps, Friedman details a process of connecting the dots, and this is no contradiction. When information is absent from memory, the mind often confabulates, but when information is present in memory, the mind can build and strengthen its associations. Thus, Friedman agrees that memory is constructive, but shows a particular type of construction that depends on available content.

Second, we construct with general knowledge. What Friedman means by “time patterns” can include natural cycles (seasons, daylight, weather, lunch), conventional timekeeping (hours, days, weeks, months, years, decades), recognized progressions (“sophomore, junior, senior”, “rookie, journeyman, vet”), and more. Friedman also discusses “temporal locations” (one particular “step” in any “time pattern” just listed) and “temporal landmarks” (epoch defining events like a wedding, a war, an election, or essentially any change deemed significant).

Third, we construct with “associations” that are currently also in memory. The context of “trace memory A” can sometimes be found within “associated memory B” (C, D, etc) and sometimes within “trace memory A” itself (through decompressing the original encoding). In both cases, that data is built together if and only if information supports temporal inferences. Friedman never discusses the possibility of inventing material because his focus is on what the mind does with a specific type of material. When subjects were asked to remember a time sequence, their ability to “succeed” depended on professing to recall that type of material.

To illustrate: Suppose a retained piece of information (we had carrot cake at the cafeteria) implies its own place in a known temporal pattern (they only serve carrot cake on Thursdays). This can render a temporal location for the event (our lunch together must have been Thursday). Obviously, this constructed memory might indeed prove to be false (we actually purchased rare leftovers on a Friday) but this fact is precisely what makes the point. With some false temporal content (cake served only on Thursday), we “successfully” produce a coherent timeline, albeit a false one. However, without that false piece of data, we’d be unable to reproduce any timeline at all. This is not to suggest that false timelines are better than no timelines, but that remembered timelines are only constructed when memories proffer temporal content.

I remember calling my wife on 9/11. She was watching “The Today Show” and our toddler son remarked “Airpwane Boom Hot” (or something like that). As it stands, that memory can both receive and provide temporal context by triggered associations. First, I happen to remember the year (2001). More significantly, this memory reminds me of a time period while Bo was little when Sarah stayed at home mornings, and a semester when my free period (for calls) was early morning. Also, working in the other direction, the temporal context of my 9/11 memory itself can be further built up by remembering the phase of my wife’s career when she worked two nights a week at a hospital in midtown Atlanta. The details in such examples may require fact checking but their historical accuracy (or their degree of distortion) isn’t what makes them fit together as sequential “events” and “time periods” in my mind.

Notice how it’s subjective reasoning that assembles these connections - based less upon the “natural logic” of realistic situations and more upon whatever the mind puts forth as associated context. Mothers don’t necessarily stay home with small children, but I seem to recall that this mother did, for a while. Likewise, we know cafeterias can change their menus, but when we think otherwise, we find a basis for constructing memory with a temporal sequence. Though Friedman’s work doesn’t observe this subjectivism in a self-critical way, it takes the common tactic of psychological experiments to accept reported memory as reported memory. In historical criticism or the philosophy of history, some would find this naive, but in cognitive science it’s the required assumption. Remembering is necessarily a subjective endeavor.

Notice also the subjectivity of discussing “time periods” and “events”! These are common phrases in Friedman’s 1993 paper and they’re used uncritically throughout. At one point he confidently states, “one of the most consistent temporal properties of our enviornment is that events have beginnings”. In fairness, he cites “concerts, sporting events, and lectures”, but the disciples of Louis O. Mink could still pounce on this in a heartbeat. Not only does Friedman say “events have beginnings”, he even calls this a property of the environment! The statement is clearly not defiant but ignorant of Mink’s rigorous philosophy. He does not refute it or cite Mink’s opponent David Carr, and this is as it should be. This kind of narratological imprecision is still good clean work in psychological research.

As a cognitive scientist, rather than a historian, Friedman is rightfully and delightfully free to write about “events” and “time periods”. He is not looking for sequence as it previously was in the actual world, but sequence as it presently is inside of somebody’s mind.

These terms have been famously problematic in narrative theory - flashpoints in the debate on whether linguistic representation can accurately reflect human activity and/or “historical experience” - but they pass by as common parlance within cognitive science. Again, this is good. Experimenters must employ whatever terms their laboratory subjects might use to describe the world as they think they remember it. Instead of questioning whether “the French Revolution” should qualify as a single “event”, a psychologist would study how others’ minds have determined that question. Instead of debating whether the actual past has/had “beginnings and endings”, cognitive research looks at beginnings and endings as they exist in the mind of someone who perceived and/or remembers the past in such ways. This demonstrates once again the sense in which concern for accuracy can be a distraction. A particular memory may or may not reflect well the actual past, but that memory absolutely reflects something about what seems preserved in the subject’s own mind. Thus, events and time periods (and even sequence itself, which is ultimately a corollary of periodization) are fair game for remembering, as much as anything else the mind reconstructs, based on information it thinks it can recall.

We can, should, and will talk about accuracy some other time. The objective verification of someone’s remembered timeline is a worthy endeavor, and quite often more possible than some would have us believe, but that’s a question of writing a story. Today we’re talking about remembering stories - biographical stories with lengthy and elaborate storylines. To do that, we must take a lesson from Friedman. He wasn’t studying timeline accuracy, but timeline construction. And in cases of remembering time periods or event sequence, all of Friedman’s work boils down to one simple formula. Mnemonic content determines its own structural context.

Or, to put that differently, temporal content self-determines its place within sequences.

By putting it this way, we’ve generated a tautology. By definition, we would never refer to content as temporal unless it directly or indirectly implies some distinct aspect of sequence. Conversely, whenever information in trace memory serves to implicate temporal sequence, we may refer to such mnemonic data as “temporal content”. This is my whole point, which we must now apply.

So, let’s get back to discussing biographical literature.

     ~~~~~~~~~~~~~~~

By exporting our new tautology into narrative theory, we can now say that without self-structuring content in some reader’s fabula (remembered version of a narrative), that particular fabula cannot possess temporal structure. Incidentally, this challenges official definitions of “fabula” as “putting the story back in chronological order” because memory is not automatically chronological (except when given the prominent and sufficient causality of a classical plot). Among narratologists, to my knowledge, only Mieke Bal writes about the potential for fabulae to be more than mere re-orderings of story content, and her rich insights on atemporal aspects of literary remembering deserve wider recognition. However, since my purpose here is to show how readers do remember chronology, we might simply say that fabulae differ for individual readers, and that a particular fabula can demonstrate story structure only when minds reconstruct stories via temporal content.

In Aristotelian theory, of course, this “temporal content’ is provided by the sufficient causality of a governing Plot. In classic literature, remembered timelines have rebuilt themselves so effortlessly that “chronology” seemed a given property of all fabulae. Without question, there is a superior elegance in the way that remembering causality preserves evidence of the passing of time. But causality is not memory's only proof of temporality, and not every bit of remembered temporal content is an instance of causality.

Friedman’s work does not mention causality, although any memory encoding causality would of course fit perfectly with all the parameters of Friedman’s constructive theory. If the informational content of a trace memory is “A caused B”, we have a clear temporal basis for constructively remembering that ordered pair as one small segment of (hypothetically) a larger timeline.

But despite what narratologists have sometimes assumed, the chronological order of a fabula does not always depend on causality. It depends on remembering any bit of temporal content, of which causality is merely one prominent type. Another type of temporal content - the kind I am presently laboring to identify in this series - might be called character development, but we can call it anything else you might like. How about “biographical temporality”?

In the lengthy and elaborate storylines of comprehensive biographies, which are stories without a succinct plot structure, our memory of the storyline can depend upon character development. This is not because we’re obsessed with people. It’s because people change in many ways that measurably evidence the passing of time.

This (finally) raises the practical question of literary material. 

Now that we understand the mnemonic construction of storylines, the next question is: What types of content are implicitly “temporal”? What are some examples of narrative material in biographical literature that exemplifies the kind of self-structuring power readers need to achieve chronological recollection? 

There's a long detailed answer to that question, but we need a break. Come back soon for a plethora of illustrations that will unpack the second half of my teaser from part one:

In focusing on Character, we find temporal structure in self-sequencing memories of human growth (biological & psychological) and personal development (social, cultural, legal, political).

Anon, then...

Recent Posts
Recent Posts Widget
"If I have ever made any valuable discoveries, it has been owing more to patient observation than to any other reason."

-- Isaac Newton