Affective blindsight: intact fear conditioning to a visual cue in a cortically blind patient Alfons O. Hamm,1 Almut I. Weike,1 Harald T. Schupp,1 Thomas Treig,3 Alexander Dressel2 and Christof Kessler2 Departments of 1Psychology and 2Neurology, Ernst- Moritz-Arndt-University of Greifswald and 3Clinic of Neurological Rehabilitation, Greifswald, Germany Correspondence to: A. O. Hamm, Department of Psychology, Ernst-Moritz-Arndt-University of Greifswald, D-17487, Germany E-mail: hamm@uni-greifswald.de Summary Blindsight refers to remarkable residual visual abilities of patients with damage to the primary visual cortex (V1). Recent studies revealed that such residual abilities do not apply only to relatively simple object discrimin- ations, but that these patients can also differentially categorize and respond to emotionally salient stimuli. The current study reports on a case of intact fear con- ditioning to a visual cue in a male patient with complete bilateral cortical blindness. The patient was admitted to the stroke unit of the neurological department because of complete loss of vision. Both CT and structural MRI scans confirmed lesions in both territories of the poster- ior cerebral artery. No visual evoked potentials could be detected confirming complete cortical blindness. During fear conditioning, a visual cue predicted the occurrence of an aversive electric shock. Acoustic star- tle probes were presented during and between the con- ditioned stimuli. Relative to the control condition, startle reflexes were substantially potentiated when eli- cited in the presence of the conditioned stimuli. No such potentiation was observed prior to conditioning. These data suggest that fear learning to visual cues does not require a cortical representation of the conditioned stimulus in the primary sensory cortex and that subcor- tical pathways are sufficient to activate the fear module in humans. Keywords: blindsight amygdala fear conditioning startle Abbreviations: CS = conditioned stimulus ITI = inter-trial interval PCA = posterior cerebral artery SCR = skin conductance response TS = test stimulus US = unconditioned stimulus VEP = visual evoked potentials Introduction Blindsight refers to remarkable residual visual abilities of patients suffering from damage to the striate cortex (V1). It has been demonstrated that these patients can accurately detect, discriminate and localize visual stimuli presented in their blind field, without being able to report any accom- panying conscious visual experience (Weiskrantz, 1997, 2000). A number of studies showed that cortically blind patients are able to behaviourally discriminate different colours, simple shapes or movements of objects while they insist that they cannot see these stimuli (Barbur et al., 1980 Stoerig and Cowey 1992 Weiskrantz, 2000). Recently, De Gelder and colleagues reported for the first time that such residual abilities do not only apply for relatively simple stimulus properties, but for emotional salience of stimuli as well (De Gelder et al., 1999). In this study, De Gelder and colleagues presented short video clips of a female face pronouncing the same sentence with either a happy, angry, sad or a fearful facial expression to a patient (G.Y.) with damage to his left occipital lobe. In various forced choice tests, patient G.Y. was able to discriminate between the different emotional expressions above chance when presented in his blind hemifield. G.Y. was not aware of the faces he responded to. It has been suggested that this remaining but non-conscious visual capacity might be mediated by an extrageniculate parallel visual pathway to the extrastriate cortex bypassing the striate cortex (V1) and involving the superior colliculus and the posterior visual thalamus (pulvinar), which remain functional in a blindsight patient. Recent neuroimaging data by Morris and colleagues suggest that these residual abilities might also be amygdala-dependent (Morris et al., 1998). In their first study with unimpaired subjects, Morris and colleagues detected stronger activation of the amygdala in PET scans in response to a conditioned stimulus (angry face) that was followed by an aversive sound. To experimentally reproduce ���blindsight��� in these sighted volunteers, conscious �� Guarantors of Brain 2003 DOI: 10.1093/brain/awg037 Brain (2003), 126, 267���275

perception of the conditioned stimuli was prevented using backward masking. In this masking condition, the angry faces were presented very briefly (30 ms) and then immediately followed by a second, masking stimulus. During processing of the reinforced conditioned stimulus (CS), a stronger activation of the right amygdala was observed in the masking condition, while in the non-masking condition the reinforced face elicited a stronger activation of the left amygdala. In a follow up study (Morris et al., 1999), it was found that activation of the right amygdala produced by the unseen CS was reliably predicted by the activation in the superior colliculus and pulvinar. In contrast, such a relationship was not obvious when the amygdala was activated by non-masked stimuli. Recently, Morris and colleagues reported that the same colliculo���pulvinar���amygdala pathway is also activated dur- ing processing of fearful facial expressions or fear condi- tioned faces when these stimuli are presented in the blind hemifield of a patient with a unilateral striate cortex damage (patient G.Y.) (Morris et al., 2001). As expected, faces that were presented in the intact (left) hemifield of G.Y. elicited enhanced activation in the intact right visual cortex (com- pared with the blind hemifield presentations), but this activation was not modulated by the emotional expression or the conditioning history of the faces. In addition, fearful facial expressions and fear conditioned faces presented in the blind (right) hemifield did not evoke increased responses in the intact striate cortex (relative to the happy facial expres- sions or non-reinforced faces), but nevertheless elicited increased activation in the amygdala and in the superior colliculus. Moreover, this differential amygdala activation showed a condition-dependent covariation with the visual thalamus and superior colliculus consistent with the involve- ment of this pathway in processing fear-relevant stimuli. Thus far, research on affective blindsight has revealed that a cortically blind patient can discriminate at above chance level between different emotionally salient stimuli and that the colliculo���pulvinar���amygdala pathway might mediate this residual ability. The present study examined whether a cortically blind patient can also acquire a reliable fear response to an unseen visual cue that is paired with an aversive event. Specifically, we investigated whether these subcortical pathways not only modulate implicit stimulus discrimination on a perceptual level, but can also shape simple reflexive behavioural adjustments to unseen fear- evoking stimuli. Animal data suggest that the acquisition and expression of a reliable fear response do not require a representation of the aversively conditioned stimuli in the primary sensory cortical areas (Falls and Davis, 1993 LeDoux, 1996). We employed the startle probe methodology in our experiment to assess fear conditioning in a cortically blind patient. The startle response���a cranial to caudal spreading wave of flexor movements along the neural axis���is a primitive protective reflex that is elicited by an abruptly occurring sensory event of certain intensity (Berg and Balaban, 1999). During fear conditioning, the induced fear state of the organism facilitates this independently instigated protective reflex (for review, see Davis, 1998). The use of the startle reflex as a measure of fear conditioning has a number of advantages. First, conditioned and unconditioned anxio- genic phenomena can be measured by the modification of a simple reflex. Secondly, the reflex per se is not a specific component of the fear state (like freezing), but rather a response to an independent probe event that is primed (facilitated) when the fear state is present. Thirdly, the reflex can be elicited by a stimulus that can be easily controlled by the experimenter. Fourthly, the neural circuitry of the fear- induced facilitation of the acoustic startle reflex is very well described (Davis, 1998). Converging evidence indicates that the amygdala with its efferent projections represents the key structure that modulates the fear potentiated startle effect. Finally, fear conditioned startle potentiation can be reliably observed in humans and has proved to be highly replicable across laboratories (Hamm et al., 1993 Hamm and Vaitl, 1996 Lipp et al., 1994). Finally, in contrast to skin conductance learning, startle potentiation occurs specifically during fear conditioning, but not during non-aversive learning Fig. 1 (A���D) CT of the patient���s brain at admission showing a pre-existing infarction in the right (R) PCA territory and a severe hypoattenuation in the left PCA territory indicating recent ischaemia. 268 A. O. Hamm et al.