Analysis of Route-Based Spatial Segregation through Rat Parietal Cortex Neuronal Discharges

by Mariya Kopynets

Final Paper submitted in Partial Fulfillment for Cogs 119 Course Requirement for Bachelor of Science degree in Cognitive Science with a Specialization in Neuroscience.
Professor: Ayse P. Saygin
Teacher’s Assistant: Carson Grant Miller Rigoli
University of California, San Diego
December, 2016


The Parietal cortex plays an extensive role in decision making in relation to one’s environment. Understanding how individual Parietal Cortex Neurons (PPC) neurons are activated can elucidate how specific decisions are made in relation to a particular spatial fragmentation in relation to frames of reference. Recordings of electrical signals from the PPC neuron in the form of action potentials in rats trained to undergo a left or right turn in a T maze exhibited that specific neurons converge or diverge depending on the location of the decision or turn movement. Most importantly, the PPC neurons found to fire at high rates prior to the motor action, confirming the role of Parietal cortex in planning and predicting, as well as route-centered positioning. The analyzed data suggest that specific neurons do fire at unique locations and it may be that these neurons can work in an ensemble network to provide stronger analysis of how decisions are processed in the Parietal cortex and beyond.


The discovery of brain modularities present with profound understanding of the function of the four major lobes of the cerebral cortex in the brain of mammals. There is however a significant amount of research to be assembled in pursuance of comprehensive understanding. The parietal lobe for example is known to be essential for motor function planning, helping species to execute the motor behavior with conscious action prediction ( Blakemore, 2003) . Charles Scott Sherrington a British neurophysiologist and Nobel prize laureate in 1941 stated that all action is a mere internalization of our thoughts. His statement lead to the discovery of parietal function in the early 20th century (Rauschecker, 2004) . Today it is generally accepted that the parietal lobe is positioned superior to the occipital lobe and posterior to the frontal lobe having profound crosstalk functions with the Hippocampus. Functions associated with neuronal activity in the Parietal cortex are the five human body senses regulation, perception, sensation, and visual input and movement coordination. Among the main functions however is the integration of sensory input from different modalities to cognitively translate such content into navigation information and spatial cues. Specifically, it coalesces information from visual system and the lateral postcentral gyrus, which allows the parietal cortex to map pathways to elucidate an egocentric environment.

The surrounding ecosystem in general could be spatially segregated into two frames of reference: egocentric frame of reference and arbitrary frame of reference. The latter is further divided into more distinct categories such as 1) object-centered, one’s position in the surrounding relative to an object, 2) route-centered frame of reference refers to one’s specific location in a route sequence, 3) allocentric or world-centered frame of reference indicate one’s location with respect to the surrounding world (Fillimon, 2015, Chu, 2016). The general knowledge states that egocentric positioning is controlled by the parietal cortex and arbitrary positioning was chiefly, but not exclusively, represented in the hippocampus. Nevertheless, the parietal cortex plays a pivotal role to understand spatial recognition and memory.

A recent fMRI study showed that imagining a movement had activated the left posterior and inferior parietal lobe. This study demonstrated that patients with a lesioned parietal cortex had an impaired ability to imagine the processes of predicting and planning the required time for accurate finger-pointing gesture. This experiment highlighted the importance of the parietal cortex for motor function execution imagery. The process of imagining the plan for action is an effortful complex mimicking of one’s motor actions. The imagery requires retrieving from memory a model of a similar motor action that was consolidated earlier into a stored model (Sirigu, 1996). Ergo the process of abstract representation can be alluded to the parietal cortex. However, investigations of the firing of neurons in the parietal cortex needs further assessment for both imagery and spatial recognition.

Another study done by Nitz and his team showed that PPC neurons were also heavily discharged to portray a route sequence (Nitz, 2006). During the experiment, if the positioning of the track where the animal performed the behavioural experiment was dislocated, matching neuronal discharges was obtained. Hence PPC neurons present action potentials for egocentric positioning and route-sequence mapping.

The goal of this paper, rather than providing answers about the differential ways the parietal cortex accomplishes solo the motor prediction, is to propose the potential of an orchestrated firing of ensemble of neuronal networks across the parietal cortex. Specifically, to accentuate the role of Parietal cortex in egocentric and route-centered positioning as well as in planning and predicting of motor functions, we will present the analyzed parietal cortex’ neuronal discharge.


In this paper we are conducting further analysis on existing research performed by Dr. Nitz, Douglas laboratory at UC San Diego. Dr. Nitz and his team investigated the parietal cortex neuronal discharge in Route-Based Spatial Segregation.

Experiment setup:
Four T-maze set ups were completed, with various object-cues:computer, chair, and table dispersed in a dimly lit room. The dimensions of the T-maze were: height 48 in., wideness 4.25 in., positioning 1.5 in. above the ground. The side dimensions of the T-maze were: height 26.38 in., wideness 4 in., positioning 1.5 in. above ground. All 4 T-maze set up had equal dimensions and were located 8 inches apart in vertical position along the same axis. The room was divided with an arbitrary line in the middle, with Locations 1 and 2 of T-maze positioned below the arbitrary line and Locations 3 and 4 T-maze positioned above the arbitrary line. Each of the 4 T-mazes were divided into equal 100 sections called Bins. At location Bin 83 the Turn was executed from a vertical into a horizontal (either Left or Right) direction. (PPC).

Figure 1: T-maze test set up in 4 positions across the room. Positions 1 and 2 T-maze is segregated with an arbitrary horizontal line from Positions 3 and 4. In Positions 1 and 2 T-maze the rodent trained to perform Left turn, in Positions 3 and 4 T-maze the rodent(s) is trained to execute a Left turn (Chu, 2016).

Rodent training:

Two male adult Sprague Dawley rats were trained to perform Right turns at Location 1 and 2 of T-maze, and to make a Right turn at Locations 3 and 4 of T-maze test. For correctly selecting the turn rodents were rewarded with Cheerios at the end of the T-shape (given half a cheerio). After the training, rodents had surgically implanted electrodes into their Posterior Parietal Cortex.

Recordings Neuronal Activity:

To translate how prediction, planning a movement, or conducting a specific movement with egocentric positioning we analyzed neuronal single unit firing. Explicitly, elec trophysiological techniques were utilized and measured the single unit neuronal recording that records the electrical signals from the neuron in the form of action potentials. These action potentials are later interpreted in affiliation to the question of the null hypothesis.


During the experiment the positioning of the track 4 location was randomly placed. After the training, both rodents performed with 90–100% accuracy, which lead to the conclusion that rodents learned the abstract spatial rule.
 The rodent behavior was recorded with Graphical User Interface (GUI) in Matlab. Desiree Chu’s Honor’s Thesis provides with extensive description of the experiment setup and how she performed the data collection (Chu, 2016).


After obtaining the already pre-processed raw data, our data file interp_out_rates_trial_pos1–4 contained values for 97 neurons with each line storing the firing rate vector across 100 bins (x-axis) for a particular neuron at one of the 4 track locations. There were many lines for each 1:97 PPC neuron representing the numerous trials performed. Each neuron contained values from 24–35 trials at 1:4 track locations respectively. The trialmap_cell_trial variables stored the neuron number (1st column), and the trial number (2nd column).

1. Our initial approach was to take the mean firing rate for each of the 1:97 neurons at all 4 trials, in order to obtain one average firing rate for each neuron across all 100 Bins at each of the 4 track locations. Once we successfully obtained the mean values the task was to eliminate the insignificant firing rates. The threshold was set at 5Hz firing rate across all Bins (columns 1:100). If the maximum firing rate for each neuron was less than 5 Hz all values at the assigned row were replaced with zero. In the following step we asked if Neuron 1:97 at Position 1 and Position 2 and Position 3 and Position 4 was equal to zero, such corresponding neuron was deleted. As a result we obtained 58 neurons that had a max firing rate across all 4 track locations to be higher than 5Hz. Plotting the 4 track positions for each neuron allowed discerning differences within each neuron to visually detect whether there were robust neuronal firing rates when the turn (Right or Left) was executed (around Bin 83), or whether there were increased or decreased firing rates to signal route-centered positioning.

2. With clean data we were also curious to visualize whether there were differences for egocentric positioning preferences; specifically, to test whether the rodent had a preference for a right or left turn execution. For each neuron we took the mean of Position 1 and Position 2. At these locations the rodent was trained to execute the Left turn. We then compared such values to the mean computed across Positions 3 and 4, where the rodent was demanded to make the Right turn.
 3. The third approach to analyze the data was to juxtapose Easy (mean across Position 1 and Position 4) versus Hard (mean across Position 2 and Position 3) to determine whether PPC neurons will have distinct firing rates for locations further away from the arbitrary line, whether it is harder for the rodent to perform the turn with accuracy near the segmentation line, or will there be no statistically significant differences proving that perhaps confounding variables served as cues for the learning purposes and accuracy in performance and not the segmentation line. Once the mean firing rate was acquired, Matlab functions were utilized to compute the probability distribution function (PDF) to determine the probability of a specific neuron firing at a specific bin location. We competed to sum of all firing rates across 100 bins for each neuron at each positions. Then the firing rate at each bin location was divided by the sum, providing with the PDF value to be stored. A repeated PDF was computed for all 3 conditions (juxtaposing 4 positions for each neuron, comparing mean Left to mean Right, contrasting mean Easy and mean Hard) at designated track locations respectively.

The values from PDF were stored and applied for computing the cumulative distribution function (CDF). CDF is the sum of probabilities of a neuron firing at a specific bin location and below. These CDF values were then applied to compute the Kolmogorov Smirnov statistical (kstest2 function) test which showed the statistically significant differences within specific neurons at a definite bin location. Besides the CDF, we also computed the slope, which allowed to determine at what rate the changes in firing occurred the fastest within each neuron for the comparison of mean Right and mean Left condition.


Figure 2: Subplots of Neuron 145 Illustrating Firing Rates in Hz (y-axis)-1st Column and in probabilities for CDF (y-axis)-2nd,3rd, 4th Column, and respective bin locations (x-axis).

Neuron 145 shows higher firing rates at bin 70 for combined locations 1 and 2 and at bin 80 for locations 3 and 4. From this data it was determined that neuron 145 was pivotal for the rat to turn left or right. Moreover, this PPC neuron predicts the Left turn, since the 1st column figures plot the Right turn (red color) prior to Bin 83 when approximately the turn was made. Hence the rodent predicted the turn and the peak action potentials confirms this finding. When segmenting the locations with the mean across positions 1 and 4 (Easy) and 2 and 3 (Hard) there was not much difference. Hence, neuron 150 seems to only play a large role in the decision making process for the non-segmented locations.

The slope showed that around bin 70 there is a sharp peak- the rate at which the difference in neuronal firing occurred.

Figure 3: Subplots of Neuron 2 Illustrating Firing Rates in Hz (y-axis)-1st Column and in probabilities for CDF (y-axis)-2nd,3rd, 4th Column, and respective bin locations (x-axis).

Neuron 2 showed high firing rates after the turn (bin 83) and a divergence in firing rates in segmented locations. Perhaps neuron 2 is important for understanding egocentric behavioral recognition, amplifying neuron’s inclination to fire for both Right and Left turns (as shown at bin 83–90).

Figure 4: Subplots of Neuron 170 Illustrating Firing Rates in Hz (y-axis) and in probabilities for CDF (y-axis) and respective bin locations (x-axis).

Neuron 170 shows a large divergence before the turn at bin 83 and has stronger firing rates and divergences. The neuron is constantly fluctuating prior to both turns. Perhaps neuron 170 is a very pivotal neuron in deciding the rodent’s egocentric positioning with a high preference for Right turn taking. The peak firing rate confirms that such neuron has a preference for taking an action.


It is interesting to note how specific neurons are activated before reaching a decision (left or right turn), but these specific activations do not correlate well with overall action planning and execution decision by the rodent. While analyzing a specific neuron which elucidates important localized areas of the brain that are necessary to map certain decisions, Dr. Nitz suggests that an ensemble network of neurons in the PPC (Shelley, 2016) may suit a better understanding of action planning and execution. Furthermore, the Parietal Cortex is not the only location of the brain important for environmental recognition, but the hippocampus plays a pivotal role as well. Dr. Nitz further suggests that a significant cross-talk of ensemble networks of neurons between the Parietal Cortex and Hippocampus (Nitz 2006) may be responsible for having a greater understanding of decision making in spatial segmentation. Therefore, instead of analyzing individual neurons firing at a maximum rate at certain bin locations, it may be beneficial to assess how unique neurons fire synergistically at different bin locations. Such data could provide with higher insight to the crosstalk that is relevant to making a decision based on egocentric and arbitrary positioning behaviors. It was decided by our team that instead of investigating the ensemble neuronal network that synergistically fired, we researched the neuronal firing prior to Bin 83 in correlation to the rodents behavior to execute the turn (either Right or Left). However, after reviewing the final data, further investigations are recommended for the direction to examine ensemble of neuronal firing at different Bin location. This will provide further insight to the questions whether PPC neurons are highly activated in route-centered positioning and object-centered positioning. Since during the training there were object-cues distributed in the experiment room, which have have provided with additional visual and sensory modalities to help rodents remembering the route sequence and the correct turn taking at the specific 1:4 T-maze locations either above or below the arbitrary line. Further investigations are recommended to see how the rodents may have not learned to make the correct turn, and it could be beneficial to determine whether object-centered positioning is represented by PPC.

From my submitted code that the team did not ultimately include for our presentation, I found there was an ensemble play of neurons firing at respective bin locations. This code took into account local maximums of each individual neuron and paired any other similar local maximums of other neurons at respective bin locations. Interestingly, there were a few neurons that simultaneously fired from analyzing the local maximums. This corroborates that there is an ensemble firing of neurons in the parietal cortex at different bin locations that could have an important factor for route-centered positioning. And such finding could ultimately be linked to determining how the rodent would make a decision. Further analyses of this data would support the findings of Tank (Harvey, 2012) who looked at calcium transients in the parietal cortex confirming these ensemble networks for a behavioral succession in an environment.

Lastly, during the data analysis process with the recommended approach to compute the mean across all trials and eliminating neurons with insignificant firing rates, we discarded 39 neurons from the 97 neurons total. These 39 neurons were firing at a computed mean rate in all 4 locations of less than 5 Hz. However, if looking at an ensemble network it is not recommended to discard neurons that are firing less significantly than others as this could bias the data. Certain neurons could be seen well activated over others mapping unique areas along route sequence (bin 1:100) and providing additional cues about the rodents’ planning and predicting behaviors for action execution. Inclusively, it would be important to revisit the analysis for not just looking at individual firing rates, but also taking all the firing rates into account and measuring local maximums for neurons that were fired during this experiment. Ultimately, this data could give a greater appreciation for the cross-talk that is crucial in understanding how neuronal networks interact in the parietal cortex to collectively make decisions in spatially segmented environments.


Blakemore, S., Action prediction in the cerebellum and in the parietal lobe, Experimental Brain Research, 153:239–245, 2003, , Accessed December 2, 2016

Chu, D, Parietal Cortex and Hippocampal Contributions to Rule Based Spatial Fragmentation, Diss. University of California San Diego. 2016. Web. Nov. 29, 2016,_desiree.pdf

Fillimon, F. “Are All Spatial Reference Frames Egocentric? Reinterpreting Evidence for Allocentric, Object-Centered, or World-Centered Reference Frames” Frontiers in Human Neuroscience, Vol. 9, no. 648, 2015, Accesed December 2, 2016

Harvey, D. “Choice-specific sequences in parietal cortex during a virtual- navigation decision task”, Nature. 2012, Vol. 484(7392): 62–68. Accessed Nov. 22, 2016

Nitz, D. “Spaces within spaces: rat parietal cortex neurons register position across three reference frames”. Nature Neuroscience, 2012, Vol. 15, 10. Accesed Nov. 22, 2016

Nitz, D., “Tracking Route Progression in the Posterior Parietal Cortex”, Neuron, Vol 49, 747–756, March 2, 2006 Accesed Nov 22, 2016

O’Keefe, J. “The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat.” Brain Research, 1971, Vol. 34, 171–175 Accessed Nov. 29, 2016

Rauschecker, A. “A Review of Parietal Lobe Functioning in Planning and Updating Motor Movements” Journal of Young Investigators, Issue 3, September 2004, , Accessed December 5, 2016

Shelley, L. et al “Creating spaces: A possible function of the hippocampal and parietal networks” Poster. Society for Neuroscience. 2016. Print

Sirigu, A., et al “The Mental Representation of Hand Movements After Parietal Cortex Damage” Science, VOL.273, 1996, , Accessed December 6, 2016