Rattlesnakes possess a sensory system specialized for the detection of infrared (IR) radiation. IR signals ascend as far as the optic tectum, where they generate a spatiotopic map. It is unknown if such signals reach the forebrain, but the existence of prominent tectothalamic pathways in other vertebrates makes this a distinct possibility. In nonmammalian forms, the major target of ascending tectal visual signals is nucleus rotundus, a thalamic nucleus that projects in turn to the subpallial telencephalon. We sought to determine whether a tecto-rotundo-telencephalic system exists in rattlesnakes and, if so, whether it carries IR as well as visual information. We have identified a thalamic nucleus in the rattlesnake Crotalus viridis that matches the n. rotundus of other reptiles in its topographic location, cytoarchitecture, and connections. Using anterograde and retrograde transport of HRP, we have demonstrated a strong ipsilateral and weaker contralateral tectorotundal projection. Tectorotundal cells lay primarily in the deeper tectal layers, which receive input from the IR system, but also in the superficial, visual layers. In n. rotundus, single units recorded extracellularly invariably responded to visual stimuli, but many were also sensitive to unimodal IR stimuli. IR and visual receptive fields were very large and often bilateral. Some rotundal units appeared sensitive to substrate vibration. Most habituated rapidly. Nucleus rotundus was found to project to a sector of the ipsilateral anterior dorsal ventricular ridge (ADVR) of the telencephalon. Single units in this region of the ADVR resembled those in rotundus, responding to visual, IR, and/or vibrational stimuli and possessing large, often bilateral receptive fields. These findings demonstrate the existence of a tecto-rotundo-telencephalic pathway in rattlesnakes and suggest that this system conveys IR as well as visual information to the forebrain. Ascending tectofugal pathways have been implicated in the discrimination of form. Thus, pattern recognition may have to be added to orientation as a proper function of the IR system of pit vipers.

The spatial and temporal characteristics of the infrared responses of single neurons in the nucleus of the lateral descending trigeminal tract (LTTD) of the rattlesnake were investigated. The LTTD is the sole projection site of trigeminal neurons that innervate the thermoreceptive pit organ. In contrast to the responses of the primary infrared neurons, which have phasic and tonic components, the neurons in the LTTD respond strictly phasically to a sustained infrared stimulus. During an excitatory stimulus, the transient burst is followed by suppression of firing or by reduction of the new rate below the rate that would have occurred in the absence of stimulation. The phasic character of the responses may enable these neurons to encode more accurately changes in the pattern of infrared stimuli. Neurons in the LTTD show adaptation within limited regions of their receptive fields, while responses in other regions remain undiminished. This indicates that each LTTD neuron receives input from a population of primary infrared neurons. LTTD neurons respond to infrared stimuli of intensity less than 0.01 mW/cm2, which is below the threshold reported for primary afferent neurons; this also suggests convergence of a number of primary infrared afferents onto each LTTD neuron. LTTD neurons have smaller excitatory receptive fields than do the primary afferent neurons in the infrared system, indicating that spatial sharpening also occurs in this nucleus. Receptive fields of LTTD neurons may have inhibitory areas flanking the excitatory area. Introduction of a stimulus into the inhibitory area results in depression of the background discharge; thus, the inhibition is due to an active process, not to rebound from excitation. Inhibition can also be demonstrated by simultaneous stimulation of the excitatory and inhibitory receptive-field areas, resulting in a decreased excitatory response. We suggest that convergence of antagonistic excitatory and inhibitory inputs can explain the time course of LTTD responses to infrared stimulation and the architecture of LTTD receptive fields. Such excitatory and inhibitory interaction, similar to that postulated for the responses of some vertebrate retinal ganglion cells, could function to provide the basis for directional selectivity, motion sensitivity, and border enhancement in the infrared system. Unlike the visual system, however, in the infrared system excitatory-inhibitory interactions allow the construction of small excitatory receptive fields in the LTTD from the larger receptive fields of the primary afferent neurons, resulting in a highly evolved trigeminal system with visionlike function.

1. An isolated eye or eye plus optic lobe preparation (in oxygenated chilled seawater) from Loligo opalescens, Octopus bimaculata, and O. bimaculoides was used to study the electroretinogram (ERG) for small signal intensity-modulated stationary spots of light. 2. If light intensity was modulated sinusoidally (modulation depth 0-50%) the ERG response is sinusoidal with less than 2% of the power present in the next five harmonics compared to the fundamental. Bode plots, amplitude and phase shift plotted against frequency, were constructed from these sinusoidal input-output experiments. 3. Linearity and time invariance were tested: a) an increase in amplitude of sinusoidal modulation by a constant factor caused an increase in response amplitude by the same factor but caused no change in shape of the Bode plot gain or phase curves; b) the transfer function represented by the Bode plot could be used to predict waveshape of the response to a brief flash (Green's or impulse-response function); c) the Fourier transformed square-wave response could be used to obtain a Bode plot which coincided with that obtained by sinusoidal input-output experiments. 4. The Bode plot can be fit by the transfer function of 5-12 (depending on conditions and on the preparations) series cascaded low-pass filters whose corner frequences are distributed between 0.2 and 40 Hz. Alternatively, 3-7 filters plus a delay of 25-130 ms fits the Bode plots equally well. The series filter model is compatible with a simply physical model consisting of cascaded chemical reactions whose forward rate constants are reciprocals of the filter time constants, whose reverse rate constants are negligible, and in which the concentration of an intermediate product controls membrane current. 5. As mean intensity is increased, the gain decreases. This effect is more pronounced at low frequencies than at high frequencies. Thus, the system is nonlinear for large intensity changes. The process of adaptation involves not only a change in gain, but a change in shape of the Bode plot, i.e., change in filter corner frequencies. In terms of the reaction chain model, this means that some rate constants change as the state of adaptation is changed.