S210

RESPONSE PROPERTIES OF SINGLE NEURONS IN THE INFEROTEMPORAL CORTEX OF THE MONKEY TO DIFFERENT FACIAL FEATURES

SHIGERU YAMANE*, SHINJI KAJ!, AND KENJI KAWANO Electroteehnical Laboratory, Umezono, Tshukubashi, Ibarakiken 305, Japan

Single neurons were recorded in the inferotemporal cortex (IT) of a monkey trained to discriminate three selected human faces from a large number of different faces (Neurosci. Res., s5, s114, 1987). We used 30 faces, a square and an ellipse as the discrimination stimuli, and also a small green spot that appeared after a stimulus disappeared to determine whether the monkey discriminated the stimulus or not. We examined 446 single neurons in the anterior and posterior banks of the superior temporal sulcus, including its fundus, and the gyrus of the IT. Of the 446 neurons, there were 86 face-responsive neurons. Most of them responded also to the green spot (41/86), and the square and/or the ellipse (24/86). These neurons were found in all regions. Out of 86 neurons, only 21 responded exclusively to the faces, and most of them (18/21) were located in the gyrus of the IT. Faces used as stimuli were quantified by measuring the distances between individual parts of the face e.g., eyes, eyebrows, nose, mouth, hair, etc. The correlation analysis between the quantified facial features and the response magnitudes gave correlation coefficients greater than 0.6 for five out of the 21 face-responsive neurons. Responses of these neurons correlated to the combined distances between the different parts of the faces. One of them showed correlated responses to the combination of the degree that the forehead above the left eye was covered with hair and the distance between the eyes and the mouth. In order to confirm that the neuron detected the combined facial features, six montaged faces made up of two faces, i.e., one face that excited and another that inhibited the neuron, were used as the stimuli. The relationships between the responses of this neuron to the montaged faces and the combined facial features were close to the regression line obtained from the correlation analysis. It is concluded that this neuron detected the combination of the degree that the forehead was covered with hair and the distance between the eyes and the mouth. The fact that the neuron detected essential features to faces confirms that the neuron belongs to face neurons. All of the face-responsive neurons that responded also to the square or the ellipse showed a correlation coefficient of less than 0.6 in our analysis.

VISUAL PROPERTIES OF NEURONS IN THE PRELUNATE VISUAL AREA.

TANAKA MASAFUMI, CREUTZFELDT OTTO DETLEV *+, Dept. of Aerospace Physiology, Res. Inst. of Environmental Medicine., Nagoya Univ., Furo-cho, Chikusa-ku, Nagoya 464, TDept. of Neurobiology, Max-Planck Inst. for biophysical chemistry, P.O. Box 2841, 3400 Gottingen, F. R. Germany.

We recorded activities from single neuron units in the prelunate gyrus of two awake monkeys

(Macaca sylvana). The monkeys were trained to fixate a point on a translucent screen during

recording sessions. Visual stimuli were projected onto the screen. The properties of 489 neurons

were divided into four groups: 1. Spectral-sensitive neurons, which made up 28% of the total

neurons. They showed, besides wavelength-selective properties, a variety of center-surround

interactions with inhibitory or disinhibitory effects on center excitation, or with inhibitory

effects from the center to the surround. Such center-surround interaction could be of opponent or

of double opponent type with respect to color. 2. Forty-three percent of the prelunate neurons

were responsive to various spatial features without spectral sensitivity. A special type within

this group was neurons which strongly responded to highly contrasting stimuli. Many of these

neurons were responsive to the internal structures of such stimuli. 3, Three percent of the

neurons were activated in connection with behavior. 4. Some of the activities of the neurons were

related to eye movement (1%). Neurons with spectral or nonspectral properties were clustered

within small, irregularly shaped patches, i - 4 mm in diameter. It is concluded that the

prelunate visual area is not a "colour area", but responds to various features of the visual

environment and relates them, through its subcortical and cortical outputs, to behavior.