The Brain’s Ways Of Categorizing Information: Association and Generalization

Neuronal systems are remarkably capable of making strong associations between paired cues (e.g., the growl of a tiger and threat). Associations between patterns of neuronal activity and specific sensory stimuli take place in all brain areas, yet for complex associations involving the integration of multiple sensory modalities more complex brain areas (e.g., amygdala) are required, with the most complex associations taking place in cortical areas. Under ideal conditions, this capacity for association allows the brain to rapidly identify threat-associated sensory information in the environment, allowing the organism to act rapidly to promote long term survival (see Phillips and LeDoux, 1992). Yet the remarkable capacity of the brain to take a specific event and generalize, particularly with regard to threatening stimuli, makes humans vulnerable to the development of ‘false’ associations and false generalizations from a specific traumatic event to other non-threatening situations. These processes are crucial to understanding memory and trauma. Associations between neuronal patterns of activity derived from specific sensory cues are matched against a ‘catalogue’ of previous experiences. For example, in an individual with a history of traumatic experience, a simple rise in heart rate induced by a non-threatening experience (e.g., exercise; see Case 5) can trigger a brainstem-mediated alarm response if in that individual’s past, the neuronal patterns of activation that occur with increased heart rate matched those associated with severe threat. The brain has stored this state memory-- and has generalized this neurophysiological pattern of activity to indicate threat. Generalization is an adaptive process. It was far preferable for the vulnerable human to be too cautious, too hypervigilant and to over-read non-verbal cues of threat. Learning the association between the growl of the sabertooth and danger should only take one experience. Not many individuals who required more than one trial to learn this had a chance to pass on their genes. Indeed, it is likely that certain sensory cues are genetically-coded to induce an alarm state – as witness the pervasive nature of phobias to snakes or the stereotyped fashion in which infants will exhibit distress at loud, sudden auditory cues. Because paired associations have been created in these regulatory and more ‘primitive’ parts of the brain, a pattern of incoming sensory information may be interpreted as ‘danger’ and acted upon in the brainstem, midbrain and thalamus milliseconds before the information gets to the cortex to be interpreted as ‘harmless.’ For a combat soldier from Vietnam, the sound of a firecracker will still elicit a ‘fear’ response (e.g., increased heart rate, startle response) even though he knows it is a firecracker. This man’s brainstem has interpreted and acted on the information before it has had a chance to get to the cortex to be interpreted in a more complex fashion. Brainstem, midbrain and limbic associative capabilities are at the heart of these automatic trauma-related “flashback” responses – emotional, motor and state memories. At each level of increasing complexity, the local associations become more complex. The associations in the brainstem are simple and categorical. Associations in the amygdala are more complex and allow interpretation of emotional signal and cues, including facial expressions -- and the intentionality they convey (threat, affiliation). Associations in the cortex are the most complex and may involve a variety of abstract elements -- associations between previously unpaired cues and various levels of ‘meaning’ can be made -- allowing abstract cognition. (INSERT FIGURE 4 about here) In post-traumatic stress disorders, associations between specific complex cues (e.g., helicopters) may become linked to the limbically-mediated emotion (anxiety). Limbic activation may result from cortically-mediated images (e.g., interpretation of a specific event as potentially threatening, or imagining a specific traumatic event). Once these limbic areas are activated, there may (or may not be) activation of lower midbrain and brainstem areas involved in the response to threat -- the efferent wing of the alarm response may or may not be activated. The degree of activation of the rest of the threat-response neurobiology residing in the midbrain and brainstem depends, to some degree, upon the ‘sensitivity’ of these systems. Indeed, it is likely that PTSD involves a sensitization of these systems to threat-related cues, internal or external (see below). A sensitizing pattern of previous activation from a traumatic experience can dramatically change the sensitivity of the brain’s alarm system (e.g., Kalivas et al., 1990). The result is a state of anxiety, even in the presence of what were originally non-threatening cues. A sensitized stress response apparatus then is likely a common etiology of trauma-related symptoms in children (see Perry et al., 1995). This is certainly the case for traumatized children where it has been demonstrated that exposure to chronic and repeated stressors literally alters a variety of brain stem related functions, including emotional and behavioral functioning (Perry. 1994; Perry, Southwick, & Giller. 1990).