Article Figures & Data
Figures
- Fig. 1.
The local bend behavior and network.A, Illustration of three segments of the leech midbody, producing a local bend (dorsal view, anterior is up). Each segment consists of five annuli, and each annulus forms a ring around the body perimeter. A touch to the lateral body wall elicits a bend directed away from the touched site. The bend results from shortening of longitudinal muscles near the touched site.B, Schematic of the body perimeter in cross section showing the convention for defining body wall location: dorsal midline (θ = 0°), ventral midline (θ = ±180°), right lateral (θ = +90°), and left lateral (θ = −90°). Shaded arrowsgive two examples of different stimulus locations; the unfilled arrow gives an example of a bend direction. C, Schematic outline of the local bend neuronal network. Two classes of mechanosensory neurons (T and P) connect to a layer of interneurons (∼30, with 17 identified so far; Lockery and Kristan, 1990b), which in turn connect to five classes of motor neurons that innervate longitudinal muscles in different regions of the body wall: dorsal (D), dorsolateral (DL), lateral (Lat), ventrolateral (VL), and ventral (V). There are also two classes of inhibitory motor neurons (not shown).
- Fig. 2.
Three methods of measuring the local bend response. In each panel an icon representing the body perimeter shows the stimulus location S (denoted by thearrow) and the measurement location θ (black rectangle). A, Video analysis. Two video frames (before and after a stimulus trial) show a dorsal view of midbody segment 10. The stimulator is mostly out of focus in thetop of the image; the stimulus location is denoted by awhite square. The unfilled circles drawn on each image define the line segments used to measure the response.B, Body wall tension. Tension is measured at one body location for a single stimulus trial; the peak tension is indicated.C, Electromyographic (EMG) analysis. Shown are EMG signals from three body locations for a single stimulus trial (S = −45°). Stimulus trace (top) shows the time intervalsE0 and E1 (in seconds) during which the EMG signals were analyzed (see Materials and Methods).
- Fig. 3.
EMG tuning curves. Normalized EMG response at four body locations is plotted against stimulus location. InA, the body perimeter icon illustrates the stimulus locations (arrows), and in all panels this icon shows the location of EMG measurement, θ (small filled rectangles). A, θ = −45°; B, θ = −90°; C, θ = −135°; D, θ = 180°. Data points show the mean ± SEM for five animals. Also shown is the cosine curve,b0 + b1cos (S − So), which best fits the mean data: A,b0 = 0.45, b1 = 0.41, So = −63°;R2 = 0.89. B,b0 = 0.48, b1 = 0.48, So = −113°;R2 = 0.97. C,b0 = 0.47, b1 = 0.47, So = −126°;R2 = 0.92. D,b0 = 0.48, b1 = 0.39, So = −167°;R2 = 0.85.
- Fig. 4.
Tension tuning curve. Filled circles show the normalized peak tension measured at θ = −90° plotted against stimulus location (each point is the mean ± SEM for six animals). Also plotted are the EMG data (unfilled squares) from Figure 3B for comparison. The cosine curve is a best fit to the peak tension data (b0 = 0.40,b1 = 0.60, So = −104°; R2 = 0.92; see Fig. 3legend).
- Fig. 5.
Measuring single-trial response. A, Ensemble data set. Plot shows normalized response (video analysis) versus measurement location on the body wall. A single trial consists of one measurement from each of the measurement locations, where the maximal response is set to one. The solid curve shows the cosine function that best fits the ensemble data (see Materials and Methods). B, Single-trial response curves. Shown are a single stimulus trial for EMG (solid squares;S = −90°) and video (unfilled circles; S = 0°) measurements, along with the respective curve fits. These trials provide examples of the best and worst fits (see Materials and Methods).
- Fig. 6.
Single-trial peak response location. Shown is the relationship between peak response location (θmax) and stimulus site (S). Mean (unfilled symbols) and single-trial data (dashes) are shown for both EMG and video measurement techniques. Note that the single-trial responses overlap, and thus the dashesappear as squares. The identity line, which is drawn for reference, also represents the predicted responses for the Continuous strategy. The prediction for the Categorical strategy is denoted by the shaded gray line.
- Fig. 7.
The basis for a model of motor output.A, Idealized responseR(S,θ). The predicted responses are shown for each behavioral strategy as a function of stimulus site (S) and body wall measurement location (θ). The response magnitude is given in gray scale, withwhite being the largest response. A tuning curve is ahorizontal line slice through this plot (i.e., responses at a single location to stimuli at all locations), and a single-trial response curve is a vertical line slice (i.e., responses at all locations for a stimulus at one location). B, Body wall innervation. Shown are innervation fields for the five classes of longitudinal motor neurons considered in the model. Right and left homologs for a given class are distinguished by differentdiagonal filled patterns. Overlapping fields within theD and V classes are shown bycross-hatching. The discretization of the body perimeter into 16 bins is indicated by the vertical lines.
- Fig. 8.
Motor output model. Given is a schematic illustration of the model used to predict motor neuron tuning. The body perimeter is shown in conventional coordinates; the 16 bins are shown by the dashes. The motor neurons innervating the right body wall are shown, with the gray lines indicating connections to a particular bin (innervation matrix). In thebox, a schematic flow of the model is provided. A stimulus at a location S results in activation of the different motor neurons, given by M(S). This motor neuron activity is passed through a threshold function. The model response Rm in a particular bin is determined by the summed activity of all the motor neurons innervating that bin (given by the innervation matrix,W).
- Fig. 9.
Predicted motor neuron tuning curves. Model tuning curves for each of the five classes of motor neurons (right side only) are shown for both behavioral strategies, Continuous (A) and Categorical (B). For the Categorical strategy (B), theD, V, and Lat motor neurons do not respond at all (i.e., their tuning curves are flat).
- Fig. 10.
Two-stimuli protocol. A, Single-trial responses for one experiment. The directed lines are vectors for which the magnitude and direction are given by the location and amplitude of the local bend EMG response to a single stimulus trial. Three different stimuli were given:S1 at location −135° (gray lines), S2 at location 135° (dotted lines), and both S1 and S2together at their respective locations (solid black lines). Note that the magnitude of the largest response toS1 was slightly truncated for clarity (see the mean value in the panel below). B, Mean values for the responses in A: S1, gray arrow; S2, dotted arrow;S1 + S2, thick black arrow. Also shown is the vector sum of the mean responses to S1 andS2 (unfilled gray arrow). C, Comparison of the location of the predicted and actual peak responses to two stimuli for both behavioral strategies. The prediction for the Continuous strategy (filled circles) is given by the vector sum of the responses to S1 and S2, and the prediction for the Categorical strategy (unfilled squares) is the average response to S1 alone (note that this response does not always correspond to the ideal of −135°). The predictions in each case are plotted against the measured mean response (six animals, three to six trials each). A point falling on the identity line corresponds to a perfect prediction.
Tables
EMG Video Stimulus site A θo Stimulus site A θo 180° −0.68 144° −45° −0.41 −72° −135° −0.40 −125° −27° −0.29 −61° −90° −0.40 −115° 0° 0.18 −9° −45° −0.90 −34° 27° −0.27 58° Shown are parameter values resulting from fits of Equation 1 to the ensemble data sets (i.e., pooled single-trial response curves) for a given stimulus site. Results from both EMG (left) and video (right) measurements are shown.
