A Roadmap to the Cerebral Cortices

Robert A. Moss, Ph.D., ABN, ABPP


I and some of my students have made the argument that the cortical column is the binary unit, or bit, involved in all cortical processing and memory storage (Moss, 2006; Moss, Hunter, Shah, & Havens, 2012). A column ranges from approximately 0.4 to 1.0 mm in diameter, being composed of several hundred minicolumns each of which contains around 100 to 200 neurons. We have suggested that oscillations (which would obviously be neuron firing rates) in the gamma range result in dynamic column formation in which only the outermost neurons are synchronized. The result is that only a fraction of a column’s neurons are committed to the columnar information bit which allows for structural integrity (e.g., resistance to damage). Moreover, in the presence of overlapping columns, it is believed this would allow for the large volume of information storage present in the cortex.
The information represented in each column and the circuits involved are referred to as cortical “dimensions” and are the keys to understanding how these can lead to higher cortical functions, such as language. The goal of this article is to explain some basic design characteristics of columns and circuits. This is necessary to fully appreciate the applied aspects of the model (Moss, 2007; 2013). Since this is written for the Neuroscience section and I have limited space, I will not spend time talking about the applied psychotherapy aspects, but hope I may be asked to do that in a later article for a more appropriate section. I will not cite the numerous studies used to support aspects of the model and refer interested readers to the four cited articles of mine for that information.

The first step in a working knowledge of cortical design is to look at the purpose of the brain. Simply put, the purpose is survival of the individual and, thereby, the species. This requires the ability to effectively interact with the external and internal environment. Thus, it is important to have an exact representation of that environment arrive at the cortex. Columnar organization has been noted in somatosensory, auditory, and visual primary cortex, such that specific columns respond to discrete stimuli. I proposed that when these basic columns activate, each sends its information outward and where the information streams cross, or intersect, a new higher-order column is activated. This means that the new higher-order receptive column represents the information of the columns that led to its formation. In relation to speech, the basic frequency columns activate at the sound of a voice. This leads to new columns that represent phonemes and syllables. The phoneme and syllable columns can in turn activate words columns. Similarly, retinotopically arranged primary visual columns activate higher-order columns representing basic geometric shapes whose output in turn activates columns of complex geometric designs. The feasibility of this seems realistic since it has been suggested all complex designs could be composed of 36 basic elements (“geons”), similar to the English language which is composed of 44 phonemes.

As each receptive/sensory column is formed, it in turn has long range projections to a similar sized area in the frontal lobe. That leads to the formation of a new frontal action column. In the example of speech, the higher-order action word columns in turn activate lower-order action phoneme columns (i.e., decoding) down to the level of the primary motor cortex where columns control the oral expression. This means primary auditory columns have associated primary motor column connections, receptive phoneme columns have associated action phoneme columns, receptive word columns have associated action word columns, and so forth. The direction and arrangement of the columnar circuit results in an information stream from the primary auditory left cortex through the phoneme columns to the word columns in the region historically called “Wernicke’s” area. The word columns in this region project in a posterio-dorsal direction (parieto-temporal junction). Those columns represent word combinations and sequences allowing appropriate syntax. The word columns of “Wernicke’s” area connect to the word columns of the region historically called “Broca’s” area. Such an arrangement explains why there is “localization” of certain functions based on the columns in a given region.

Another cortical dimension is sequential–simultaneous. Auditory information is sequential such that it arrives to the ear in waves and is transmitted in a temporal order. When primary auditory columns are activated in temporal order, the efferent information is conveyed in the same temporal order, or sequence. For a given higher-order column to activate, its lower order column must be activated in the correct sequence. For example, the “but” column and the “ter” column activated in that sequence activate the “butter” column. However, if reversed such that the “ter” column is activated first, it activates a “terbut” column at a different location and makes no sense based on learned language. At more complex levels, all time-based or episodic processing requires sequential processing. In like manner, when there is only one correct stimulus (e.g., recognizing my grandmother’s face), then sequential processing is involved. Since the temporal lobe involves sequential processing, it should be no surprise that episodic and association memories involve this area. Similarly, it should be no surprise that the anterior temporal lobe involves recognizing my grandmother’s face. In each of these cases, there is theoretically a single column that represents all the combined elements (e.g., the component facial features columns leading to the “grandmother” visual column or, in the case of association memories, multiple columns form other areas projecting to a single “association” memory column). Since spoken language is sequential in nature, it is also logically located along the ventral brain area.

Primary tactile input (i.e., multiple sensations arriving at the same time) involves simultaneous processing. In this case, multiple lower-order columns activate at the same time and in turn activate a higher-order column representing input from all of those columns. This pattern is thought to lead to the dorsal columns of the cortex being involved in simultaneous receptive and action processing. Notably, vision can involve both sequential and simultaneous processing. Visual information moving dorsally toward the parietal cortex has been referred to as the “where” information stream. This requires an awareness of many areas in the visual field simultaneously. Information moving ventrally toward the temporal lobe has been called the “what” stream and involves the aforementioned sequential mode. Visual information moving toward the temporo-parietal junction would involve both simultaneous and sequential processing, such as involved in relational speech and calculations. Motion detection is another function requiring both operation modes and involves area MT. In all of these cases, the patterns of sequential and simultaneous theoretically refer to the columnar processing mode involved in the information stream, or circuit.

Another important dimension is that of internal–external coding. The medial cortical columns are involved with internal and self-referential coding while the lateral cortical columns code for external stimuli. For example, the anterior cingulate area involves action columns which can be influenced by the mesolimbic-dopamanergic pathway. This connection serves to explain how this appetitive reward system leads to increased activity. Similarly, autobiographical memories are typically associated with posterior medial cortical activity. Any cognitive neuroscience text will give numerous experiments in which interactions with external stimuli lead to lateral cortical activity. Transitional regions such as the insula and temporal pole code for a combination of internal and external stimuli.

Another notable distinction is the proximal–distal to body dimension. This is in relation to the pre- and post-central sulcus cortex involving coding for primary motor and somatosensory functions. The further away from that central region the more distal-to-body the functions become. In the posterior direction vision and audition can involve stimuli distant to the body. Rostrally the columns involve non-body action functions such as problem solving abilities.

A final important dimension is global–analytical. Each of the previously discussed dimensions exists within each of the hemispheres. Right hemisphere columns follow a global processing pattern such that there are fewer total columns from the time of sensory input to the behavioral response. Having fewer total columns means that there can be a quicker response and more rapid memory formation. However, the sensory analysis and behavioral responses lack fine detail. In contrast, the detailed but slower left hemisphere processing is the result of many more columns being involved in the circuits.

The cortical dimensions outlined give a basic “roadmap” to the types of information processed in various cortical regions with the columns used in original processing being the same ones that are the stored “memories.” In this case, not only is sensory information processed by progressively higher-order columns in the areas around the primary receiving areas, but those same higher-order columns are the ones activated when recalling the sensory experience. The same holds true for motor activity in the frontal lobes. Therefore, if you imagine using a hammer, the frontal action columns involved in hand control and the parietal tactile columns involved in the sensation of holding the hammer are activated, in addition to object recognition of the anterior temporal lobe and occipito-parietal spatial visual imaging. This indicates the term “mirror neurons” simply refers to the action columns involved in “imagining” (motor planning) performing the activity without the actual movement which involves the primary motor columns. If a multi-sensory traumatic memory is activated, the columns in each of the involved lobes activate as do the medial temporal cortex columns involved in the association memory that links the columns in those lobes.

Both attention and top-down processing involve action columns controlling the posterior receptive columns. Therefore, if one is looking for some external stimulus in a general spatial location, the frontal eye fields are the locations of the action columns involved in controlling the lateral occipito-parietal columns where the anticipated stimulus will be visually processed. This has been referred to as the fronto-parietal attention system. If one is doing a working memory task, such as repeating a string of digits backwards, this involves the dorso-lateral frontal columns controlling the temporo-parietal number columns. On the other hand, if one is attending to internal or self-referential stimuli, the medial frontal columns would be controlling the medial receptive columns of the parietal lobes. Whichever, frontal columns are involved in an ongoing task are the ones that assume control at that point.

Hopefully it is becoming obvious that our cortical processing is not designed as a uniform whole. There are multiple information streams being processed concurrently within each hemisphere, with the stream and hemisphere most relevant to what is happening (e.g., where one’s attention and response is focused) being used in any ongoing response. Although we have the perception often that our “minds” are unified wholes, this is only a misperception (largely by our verbal interpreters) based on the fact that all the information is being processed in fractions of seconds.

I have not discussed the interplay of cortical regions with subcortical structures. These also have to be considered in relation to our internal and external reactions, as well as their role in creating increased arousal that leads to enhanced cortical memory storage. I have also not discussed hippocampal and thalamic cells within the context of cortical memory formation. Thus, please do not take the discussion of cortical processing as suggesting this is the only important consideration as related to psychopathology and psychological treatment. However, I hope this discussion has served as a primer for better understanding the basis of the Dimensional-Systems Model.


Moss, R. A. (2006). Of bits and logic: Cortical columns in learning and memory. The Journal of Mind and Behavior, 27, 215-246.

Moss, R. A. (2007). Negative emotional memories in clinical practice: Theoretical considerations. Journal of Psychotherapy Integration, 17, 209-224.

Moss, R. A., Hunter, B. P., Shah, D., & Havens, T. (2012). A theory of hemispheric specialization based on cortical columns. Journal of Mind and Behavior, 33, 141-172.

Moss, R. A. (2013). Psychotherapy and the brain: The dimensional systems model and clinical biopsychology. Journal of Mind and Behavior, 34, 63-89.


Cite this article:

Moss, R. A., (2013). A roadmap to the cerebral cortices. The Neuropsychotherapist, 2, 114-117. doi: 10.12744/tnpt(2)114-117


Not a subscriber? You can get all the individual articles here (members use your coupon code [wlm_private “NPT Basic|3 Year Subscription|Standard Membership|Staff|NPT Premium|Standard Monthly|2 Year Subscription”]sopq27483694jh [/wlm_private])to gain free access:

About The Author: