Can Connectionist Models Be Applied to Hypnosis
I would not be writing this article if I thought the answer was ‘no’. Connectionist models of the mind explain how language gives rise to consciousness. It shows how external stimulus can activate certain brain patterns, somewhat randomly, which are then given meaning in a linguistical output. The nodes which are activated are dependent upon both the input, and subsequent feedback given regarding the linguistical output. For example, if I see an animal with wings, two legs, and a beak, a series of neural networks are activated until and my output will be “bird”. This is either confirmed or contradicted by my experience, or feedback from others. If it is confirmed, the networks that were stimulated to achieve the output of “bird” are strengthened, and that network continues to grow until it is associated with other qualities, such as the animal flies, has feathers, eats worms, etc. Because this network is developed, if I were to see a penguin, I would come to the conclusion that this animal with wings, two legs, and a beak is a bird, and therefore must fly. This would be incorrect, and the feedback I get that this bird does not fly, but instead swims, would adjust the weights of the nodes so that sometimes when I see a bird, I might have an output of “swims” rather than “flies”. I would also develop more specific ideas of birds, less prototypical, as I develop concepts of sparrows, penguins, canaries, etc.
So what does this have to do with hypnosis? As our realities are shaped by our language, and our language (and thus meanings) are shaped by the statistical activation of our neural networks, hypnosis must work within this system to accomplish change. I believe hypnosis comes into play in three ways. The first is through the feedback mechanism. When a client, say a smoker, has a trigger which activates a neural network for the need to smoke, that need is reaffirmed through the quenching of the craving and the physiological response which signifies that the smoking output was correct. Hypnosis needs to change that feedback to make it incorrect. It needs to tell the system that given the same inputs, it needs to find a different output to be correct. The weights in the system need to change so that a healthy output is achieved, and that healthy output needs the proper feedback to reinforce the network system. In a sense it turns our notion of “bird” into “penguin” with a strong feedback loop. With the proper methods, hypnosis can provide a strong enough feedback mechanism to drastically change the statistical weights of the nodal system.
The second way in which hypnosis works in a connectionist framework is that it reshapes the input to begin with. Instead of a certain reality activating a specific set of input nodes, it begins by activating a different set of initial nodes altogether, thereby altering the subsequent network of nodes which are activated. To use the example above, instead of activating the nodes of “animal, wings, two legs, beak” to get the output of “bird” hypnosis can change the focus to instead activate a different set of nodes, such as “small, tail feathers, sings” for example. This has two effects. The most important is that it activates an entirely different neural network, which is not well-entrenched and well-developed. It may or may not lead to the same output of “bird”, but being an infantile neural network it is malleable with feedback as opposed to the original well-established one in which feedback will play only a small role. Secondly, as mentioned, the output may or may not change. It is possible that with the new set of input nodes that the same output of “bird” could be achieved, but more likely the output itself will change to something such as “robin”. Both of these effects can have profound impacts to accomplish the goals you are trying to achieve. In smoking for example, the attention on the triggers which cause the smoking behavior can be altered so that the triggers themselves activate a different neural network which is malleable and could have a different output. Instead of focusing on needing to finish a meal with a cigarette, the client would instead notice how much energy the meal gave them, which will lead to the need to take a walk to complete the meal rather than a cigarette. Should a cigarette still be the output, the network which gets there is weaker and open to feedback which can easily change it.
Hypnosis is further effective in connectionist frameworks in that it can alter the statitistical weights within the network through suggestion. So while the input nodes may be the same, they end up producing a different output than they normally would.