Abstract
This thesis examined the causal role of basolateral amygdala calcium calmodulin
kinase-IIα-positive principal neurons (BLACaMKIIα) in predictive fear learning in the rat. The
first set of experiments assessed the cell type specificity and functionality of a
chemogenetic approach (designer receptors exclusively activated by designer drugs;
DREADDs) in BLA neurons. Experiment 1a showed that DREADDs could be used to
selectively target BLACaMKIIα. Experiments 1b and 1c showed that activity of BLACaMKIIα
expressing an excitatory DREADD could be remotely controlled by CNO treatment, in vivo
and in vitro. The second set of experiments assessed chemogenetic excitation of
BLACaMKIIα on positive prediction error. Experiments 2a and 2b showed that chemogenetic
excitation of BLACaMKIIα was sufficient to attenuate associative blocking of fear, such that
fear learning was restored to a CS that would not otherwise be learned about. Experiment
2c showed that this same manipulation was not sufficient to influence simple fear
acquisition. The third set of experiments assessed chemogenetic excitation of BLACaMKIIα
on negative prediction error. Experiments 3a and 3b showed that chemogenetic excitation
of BLACaMKIIα was sufficient to attenuate overexpectation of fear, such that CSs that would
normally undergo fear loss failed to do so. While the second and third set of experiments
suggest that BLACaMKIIα are causal to aversive prediction errors, they are not immediately
informative about how this error signal is supported in BLACaMKIIα. The final set of
experiments assessed whether BLACaMKIIα support a bidirectional error signal during fear
learning using optogenetics. Experiments 4a, 4b, and 4c showed that optogenetic inhibition
of BLACaMKIIα at the point of the US was sufficient to impair acquisition and augment
vi
extinction of fear learning. However, this manipulation was not sufficient to condition
inhibition to a neutral CS. These data are interpreted to mean that increments and
decrements in fear learning, driven by positive and negative prediction error, are linked to
increases and decreases in US evoked BLACaMKIIα, respectively. Taken together, these data
provide support for a bidirectional prediction error signal instructing fear learning in
BLACaMKIIα.