Tricyclic Differential Engine - Recursive (TDE-R)

The brain has a cerebro-cerebellar architecture

What does Cerebro-Cerebellar mean? It means there are two separate parts of the brain, with two separate roles. These two separate parts are:-

1. the cerebrum (specialised for conscious and sub-conscious thought)

2. the cerebellum (specialised for automatic / unconscious thought)

The historical confusion (a word that appears much too often)  arises from the fact that these two parts of the brain must, in many ways, be functionally redundant. There is a seminal paper* by Schneider & Schiffrin (1977) which demonstrates the true nature of this redundancy. S&S conclude that the conscious mind processes information serially, but is limited in the number of chunks it can process at any one time, while the unconscious mind processes information concurrently (in parallel), and suffers no such process number limit.

Everyone has experienced the 'autopilot' phenomenon, for example when driving to work. You have driven this route so many times, and usually listen on the radio program to hear the news or relax to music, that you hardly ever remember the individual events along the way. Your sensori-motor (behavioural) apparatus has been 'hi-jacked' by your cerebellum, a 'robotic'. essentially emotionless** (and therefore unconscious) biocomputer. Consequently, you remember little or nothing, because emotional arousal is an essential pre-requisite of most permanent memory formation.

However, when you encounter novel situations, or situations whose relevance level must remain at significant levels, your (conscious) cerebrum is the one that is engaged. Thus, your long-term memory is kept open to the possible formation of permanent template patterns which may later prove useful or vital, by virtue of your mind being kept in this mode. 

Consider the 'Stroop' test (as featured in brain imaging experiment**) in which colored fonts form words which are printed on slides. For example, the word 'GREEN' will be printed in a RED typeface, and displayed for only a short time. The required answer from the subject is "red", not "green". However, the subject's automatic reflexive response is to say "green", ie to just read the word out as is. 


In TDE theory (my theory of mind, which is named after its basic computational type) the conscious brain (cerebrum) must make a voluntary judgement to SUPPRESS the automated, well-learned reflexive response produced by the unconscious brain (cerebellum). In Stocco et al's paper (co-authored by the inventors of ACT-R, LeBiere and Anderson) there is no such clear idea of the functional split between cerebrum and cerebellum. In fact, the word 'cerebellum' does not occur in its text. Yet the paper makes no large error of data interpretation- it (correctly) states that " a specific network of regions (including the left rostral prefrontal cortex, the caudate nucleus, and the bilateral posterior parietal cortices) is responsible for translating rules into executable form". Since the phenomenon dealt with is linguistic, it is the left cerebrum which has been analysed.  However, like many if not most recent experiments of this type, it simply fails to include the cerebellum in its purview.

The most fascinating part of the cerebro-cerebellar model is what it says about our relationship to the rest of the animal kingdom. We primates only have so much 'buffer' or 'working' memory available. It is true that primates are able to make it work for us in a more efficient way than other animals - we have two working memories in each cerebral hemisphere, the parietal lobe (spatial buffer) and frontal lobe (temporal(1)*** buffer), and the patterns we detect in one form are almost instantly available to the other. This means memorized spatial patterns (eg objects) are also available as memorized temporal patterns (eg processes), therefore making analogical thinking easier.

But in as many ways as we are like our primate cousins, we are unlike them. We build mega-structures like termites, we swarm like ants, we build complex ritualistic nests like bower birds, and we can guide each other to remote locations with signs like bees. None of these abilities are exhibited by the primates. They only occur in 'lower' animals which rely heavily on automatised, instinctual memory.

Consider our cerebellum (and its ancillary machinery, like the basal ganglia and the ascending and descending Reticular Activating Systems or ADRAS) as our equivalent of an ancient instinctual framework. No other primate has a similarly sized or interconnected cerebellum. While our cerebrum has a severely limited 'chunk' capacity (see Miller's Magic number 7), the memory capacity of a suitably trained cerebellum is effectively infinite. There are many people that have memorised entire books, and these people are not genetic freaks, just individuals with some extra time on their hands.

Far from lower animals being emotionless 'automata', as Descartes thought, it is in reality we humans who are closest in nature to robots. We alone amongst the animals have a symbolic computer**** nestled underneath our forebrains, capable of infinite machine-like precision


* Shiffrin, R., and Schneider, W. (1977). Controlled and automatic human information processing: II. Perceptual learning, automatic attending, and a general theory. Psychol. Rev. 84: 127-190.

** Stocco, A., Lebiere, C., O’Reilly, R.C. & Anderson, J.R.   'Distinct contributions of the caudate nucleus, rostral prefrontal cortex, and parietal cortex to the execution of instructed tasks'
Psychonomic Society, Inc. 2012

*** temporal(1)- pertaining to time, NOT temporal(2)- pertaining to the temples

**** the correct terminology is 'production system' as in Anderson & LeBiere's ACT-R

 ------------------------------ Copyright 2013 Charles Dyer------------------------------