In this essay, I describe a free verbalisation technique that helps to effectively integrate psychological issues and to achieve a more thorough understanding of theoretical positions. In the second part, I attempt to give an account of how and why the technique works within the Predictive Coding Framework.
Yet again, I’d like to tell you about a technique which has been of much benefit to me over the course of the last few weeks. This technique serves two functions. First, it can help you work through psychological and emotional issues, and second, it can be used to thoroughly understand theoretical concepts. The technique itself is surprisingly simple as it consists merely of talking to yourself out loud. Granted, this might strike you as odd because talking to oneself is not usually considered to be a hallmark of psychological sanity.
However, the idea will perhaps become less weird for you if you consider that this is what you are doing already anyway, the difference being that you do it silently in the privacy of your own mind. By this, I am referring to the constant mind chatter, the voice in your head that utters your internal stream of thoughts which is part of your subjective experience in almost every instance, barring certain deep meditation states and maybe sleep. Now, I claim that the issues and topics your mind feels drawn towards in mental chatter are not random. Rather, your thinking mind focuses on issues that are in need of further processing. If you’ve ever come home from work and noticed that your mind is still occupied with workplace problems, you know what I’m talking about. Likewise, when you’ve had a dispute with a loved one and find yourself going over the exchange again and again, this is exactly the kind of phenomenon I’m referring to. In other words, there is a logic behind the content of your train of thoughts. The technique I’m proposing to you in this text aims to make this process a lot more efficient because, as you put your thoughts into words, you process their content much more deeply and more quickly. In the following, I’ll first lay out the detailed instruction of how to do the exercise itself and then give you the best explanation I currently manage to come up with on how it works and why.
The Instruction
The most important aspect of this technique is to not censor yourself, i.e., to just express anything that comes to mind without passing judgement or attempting to control it. Your thoughts might strike you as boring, trivial, and cliché, you might feel that you’ve gone over the issue too many times already so it probably won’t make a difference this time around, you might think that your thoughts are irrationally judgemental and unfair towards others and that they don’t reflect what you really think. The trick is to say them out loud regardless. Furthermore, don’t worry too much if you’re under the impression that you haven’t hit the nail on the head quite yet with your formulation. There is no shame in starting the sentence over or proceeding to the next thought. If some psychological material needs more processing, you can rely on your mind bringing it up again at a later point in time and it will do this over and over until you’ve fully talked it through.
Likewise, try not to censor the way you speak. It doesn’t matter if your sentences are kind of cumbersome or sound stupid, the more you process your thoughts the more eloquently you’ll be able to talk about them. On a related note, try not to inhibit the flow of thought and speech too much by searching for the perfect expression. There is no room for aesthetical considerations or perfectionism in this practice except as the objects of your verbalising process. By this I mean that when you notice that you’re trying to come up with a more poetic way to express some emotional issue, just say something along the lines of “And now I notice that I’m trying to come up with the perfect way of saying it, but what really grinds my gears about his behaviour is that…” If in doubt, say it out loud.
You can do this exercise sitting or laying down with your eyes closed or, my personal favourite, walking in the woods. You can whisper or talk out loud or, if you don’t want to strain your vocal cords excessively, content yourself with articulating the words with your mouth. Just make sure you keep doing the exercise and don’t transition to merely thinking about the issue.
A more structured approach
After you’ve gained some experience with the technique, you can also use it in a more structured way and aim your focus on a particular issue which you know needs further processing. Perhaps you’ve noticed a dysfunctional pattern in one of your personal relations which, though not urgent, might call for clarification. Or maybe you’ve come across a theoretical argument in a book or a discussion with a friend that you’d like to understand more thoroughly. Personally, I’ve recently started to note down whenever I notice that any issue I come across over the course of my day seems to be in need of further processing so that I can go over it again when I find the time.
The way to make the approach more structured is simple. You start thinking about the issue and verbalising your thoughts and then let your mind follow the chain of associations in the usual way. Whenever you think that your thoughts have strayed away too far from the intended topic, just gently bring our attention back and start over. This feels a lot like bringing your mind back to the object of attention in concentration meditation. I’ve found it useful to not intervene too quickly as it might not be immediately evident how a certain association is related to the intended issue, but one develops a feeling over time for when to return.
This more structured approach might be called for in the case of emotionally charged topics towards which our mind usually has an ambivalent attitude. On the one hand, it knows that the topic is in urgent need of clarification, so it brings it up constantly, but on the other hand, it feels so extremely unpleasant to think about the topic, that the mind takes the first opportunity to think about something else. This is what accounts for the slippery quality of emotional thoughts. You can help your mind along by setting a boundary on the range of topics which it is allowed to associate about and bring it back if need be.
Another application of the more structured approach concerns more theoretical issues. In this regard, the technique I’m proposing is tightly related to the so-called Feynman Learning Technique, named after famous physicist Richard Feynman. He maintained that you haven’t understood a certain concept or problem thoroughly until you’re able to explain it in simple terms and without presupposing any technical jargon. The original technique is done in writing, but it works equally well using the self-talk method I’ve described above. You start by formulating a certain question (“Why does the Ganzfeld-method produce hallucinations?”, “Why does the heredity of a trait vary across populations?”, “Why does the meaning of a word consist in its use?” etc.) And then, you imagine how you would explain the issue to an intelligent and interested person who doesn’t have any prior knowledge and is unfamiliar with any of the concepts in question.
It used to happen often to me that I had read something interesting in a book and thought that I had understood it only to find that I had no clear grasp on the issue when I tried to tell a friend about it. This sort of experience clearly demonstrates that you need to put a new thought into your own words over and over again until you’ve completely assimilated it into your web of beliefs. This technique is also the perfect antidote to merely accepting beliefs that are universally held in your circle of friends. Particular intellectual circles are so homogeneous in their beliefs that certain positions and concepts can be presupposed in all conversations and never have to be defended against criticism and be subjected to careful scrutiny. For example, on two different occasions I’ve heard people profess the belief that everything is political and then, as soon as I had asked them how they understand the notion “political”, they could provide no further explanation of what they meant.
The cognitive mechanism
Now, why should talking to yourself out loud help you work through your psychological issues and to clarify your intellectual understanding? Let my start by pointing out that when you talk to yourself, you’re not learning anything that you hadn’t already known beforehand. Yet, you might come to new conclusions based on your prior knowledge and you might see familiar issues in a new light. I’ve found it to be a great source of learning, but what do you learn when you don’t get any new information?
I will attempt to provide an answer in broad strokes within the Predictive Coding Framework. To start in the most general terms, your brain is charged with the processing of information which is either taken in from the environment in perception or issued to the environment in behaviour. According to Predictive Coding, the process of perception is not one of completely passive reception of outward stimuli which the brain then assembles to form a picture of the environment. Rather, the brain also uses its past experience to make predictions about how the object of perception will appear next.
For example, if you’ve just seen a ball roll behind a box, you’ll expect it to come out one the other side. This expectation is based on your past knowledge about movement and object permanence, i.e., the belief that objects keep existing even if you don’t perceive them anymore. The expectation you form is then revised in the light of the actual sensory input. You might have expected the ball to come out on the other side at roughly the same speed but now you find it to be considerably slower. You conclude that the texture of the floor behind the box is rougher and has slowed it down (which would be another prediction). According to Predictive Coding, all instances of perception contain these two elements: the brain makes a top-down prediction about what sensory input it will receive next and then revises this prediction in light of the bottom-up sensory input provided by the sensory organs. Then the cycle starts over, the recently revised prediction is used as the new starting point, only to be revised yet again, this time in the light of new sensory input. The process is thus iterative in nature.
As I’ve described the inner workings of the brain thus far, I made it misleadingly sound like there are two levels of processing: level A that is responsible for the generation of predictions and level B that is responsible for receiving the sensory input. B then receives A’s prediction, verifies whether it matches the actual sensory input and then generates a prediction error signal. By sending this signal, B communicates back to A how far off the original prediction was from the sensory input. Level A then uses the prediction error signal as feedback to revise its prediction and then sends down the revised version so the process starts over.
This crude picture needs to be complicated further by introducing the idea of a processing hierarchy. This means that there are not only two, but rather a multitude of processing levels stacked one onto the other. Each of these levels sends down a prediction about what’s going on in the processing level below it and receives feedback in the form of a prediction error signal which is then used for revising the original prediction. Furthermore, the same level of processing receives a top-down prediction from the level above it, compares it to its own current activity and then generates a prediction error that it sends back up. As one moves up the hierarchy, the predictions become increasingly complex. Whereas the lowest level might be concerned with predictions about the smallest units of perceptual input (the mind pixels, if you will), the levels above are concerned with shapes and lines and the levels above these make predictions about objects such as cats on mats and faces and so on until one reaches the top of the hierarchy to find processing levels concerned with the Communist revolution and a sense of personal identity.
Crucially, not all predictions are perfectly revised at all times and in all positions of the hierarchy. Taking the information provided by the world into consideration is a costly process so the brain relies on its prior knowledge whenever it can. Its goal is not to make all predictions as precise as possible under all circumstances, but to allot the scarce resources it presides over to the prediction updating processes that are most urgent.
But there is another reason why it can make sense for the brain to rely on prior knowledge. The sensory information the brain receives in perception isn’t always equally trustworthy, which is to say, it is not always equally useful for the task of constructing an accurate representation of the world. This is because any sensory input consists both of signal and noise, where the signal is that portion of the input that informs us about the object of perception whereas the noise is due too random variations that are useless for representation purposes. To illustrate this distinction, think of the relation of signal to noise in a radio program. Now, it is most important for the brain to know how trustworthy a given sensory input is because, ideally, the revision of a prediction is based on perceptions with a strong signal and little noise. Imagine you want to assess whether the singing voice of a new acquaintance is really as beautiful as everyone claims. You could easily confirm or discard this prediction in the good acoustic conditions of a silent room, but less easily on the dance floor of a loud party. In revising its predictions, therefore, the brain factors in how reliable the received input is. The technical term for this estimate of the ratio of signal to noise is “precision”.
If a certain level of processing is estimated to be low in precision, the level above it will inhibit its prediction error signal. Coming back to the simpler picture with only two processing levels, if level A sends a prediction down to level B, but also sends down the additional prediction that B’s precision is going to be low (perhaps because it knows that auditory input in unreliable on dance floors), then B will inhibit itself and produce a much weaker prediction error signal. Consequently, level A will receive a weaker signal and update its prediction to a lesser degree. From this mechanism it follows that top-down processes, i.e., reliance on prior knowledge, is more pronounced in situations with low precision.
Against the backdrop of these considerations, an answer to our original question begins to emerge. This question was thus: how can it be that we learn and come to new conclusions through the self-talk technique although we don’t gather any new information about the world? If no information is transmitted from the world to the brain, then it must be a transmission between parts of the brain and therefore, plausibly, between different levels of the processing hierarchy. We have seen above that not all predictions are always revised to the extent they that could because the brain prudently budgets its resources and focuses only on the most important prediction errors. This could explain why newly adopted theoretical views can sometimes seem to exist in utter isolation from the rest of our prior knowledge. In the same vein, the whole of the philosophical endeavour might be characterised by the attempt to integrate our theoretical views into a coherent web of belief. (Note that this is similar to Wittgenstein’s view on which philosophy is the gathering of memories for a particular purpose.)
From a Freudian perspective, one might expect that the flow of information between processing levels is further impeded by the forces of psychological defense as some memories and perceptions pose a threat to our self-image and are therefore denied or suppressed. If this is true, the cognitive mechanism behind the Freudian unconscious probably amounts to a permanent attribution of low precision to the neural processing units that represent the memory in question (Peter Kuhn’s idea).
Now, how does the verbalising process fit into this picture? Is there a way of increasing precision such that the predictions of different processing levels are updated to the extent the prediction error demands it? It turns out, there is. Consider what happens when you pay attention to a certain perception. The level of detail increases immediately such that you become aware of all the minute particulars and temporal changes. In consideration of what was said above, this is an indicator of an increase in bottom-up information processing, which in turn points to an increase in precision. On this view, therefore, attention is nothing but a voluntary increase of precision. Lastly, consider that verbalising a thought requires you to pay much closer attention to it than you would otherwise, and the picture is complete.
In conclusion, we’re left with the following account. In mental chatter, the mind brings certain isolated psychological and theoretical representations to attention that are in need of integration into the cognitive hierarchy. They are isolated because the processing units which represent them are attributed low precision either to save resources or because the conflict between these representations and the person’s self-image is too painful. Thinking about these issues is an ineffective process of integration and can be much enhanced when the issues are verbalised again and again because verbalisation presupposes attention which is tantamount to an increase in precision and leads to a stronger prediction error signal and an appropriate updating of higher-level predictions.
Outlook
This account is lacking in a crucial regard because it has presented only half of the Predictive Coding Framework as it has limited its discussion to the topic of perception, leaving behaviour and their interplay completely out of the picture. Obviously, speech production is a form of behaviour, and the verbalising process therefore cannot be completely understood without taking the whole framework into consideration. At the present moment, however, I do not know enough about the role of language in Predictive Coding and will therefore postpone the discussion of this aspect to the future. Note that a complete explanation of the efficacy of self-talk will at the same time explain the positive effects of self-expression in psychotherapy. Indeed, Carl Rogers, whose client-centred approach consists mostly of actively listening to what the patient says and repeating it back to her in the form of paraphrases, often writes about the human psyche in a way very much in accordance with the conception of the human brain as a predictive processing hierarchy. The self-talk exercise is also extremely closely related to the technique of free association used in psychotherapy. But those are topics for another time.
Let me just finish with the remark that I find the beneficial effects of the self-talk technique in my own life extremely encouraging and empowering because it represents to me a self-therapeutic tool whose accessibility is not restricted by any external institution, but which is available anywhere at almost any time. If this should turn out to be a useful and effective exercise to achieve psychological integration for a certain percentage of the population without the need of a trained professional, this might be a way to satisfy the high demand at least partly for psychological support which still outstrips the corresponding offer by a large margin. If you try out this technique and find it beneficial (or not), get in touch!
Comments
Post a Comment