concurring.
I am constrained to agree that the evidence was sufficient to convict Rivas, the front-seat passenger, of the offense of possession with intent to distribute, based on a theory of constructive possession.1 I reach that conclusion reluctantly, because the only evidence upon which the government relied to implicate Rivas is that he had been seated in the front passenger seat next to a center console upon which the police discovered two ziplock bags of cocaine while Rivas was outside of the car speaking to an acquaintance on the sidewalk several feet away from the car. There was no evidence presented on the length of time that Rivas had been in the car or whether Rivas was engaged in an ongoing drug venture with Melgar, the driver, or with the two passengers in the backseat.2 The fact that the drugs were close enough to Rivas to have been in his plain view and accessible to him is sufficient to establish the first two elements of constructive possession — knowledge of the drugs and the ability to exercise control over them. See Brown v. United States, 546 A.2d 390, 394-95 (D.C.1988). Language in our cases unequivocally states that the same evidence of proximity of contraband in plain view also is sufficient to establish the third element necessary to a showing of constructive possession — that Rivas in some way controlled or had the intent to exercise control over the drugs. See Bernard v. United States, 575 A.2d 1191, 1195 (D.C.1990) (citing Brown, supra, 546 A.2d at 394 n. 2) (explaining that to establish constructive possession, the government must prove that the appellant “knowingly had both the power and the intention at a given time to exercise dominion or control”); In re T.M., 577 A.2d 1149, 1151 (D.C.1990) (same).
Although we have, as the majority notes, said that intent to exercise dominion and control “may be inferred from the presence of contraband in an automobile, in plain view, conveniently accessible to the defendant,” see Burnette v. United States, 600 A.2d 1082, 1083 (D.C.1991) (per curiam), I question whether that statement either makes sense or is supported by a close reading of our cases. First, the cases in which we have concluded that the evidence is sufficient to establish constructive possession have had additional facts supporting that the defendant had the requisite intent. See Parker v. United States, 601 A.2d 45 (D.C.1991) (concluding evidence sufficient for constructive possession where evidence showed that “narcotics-laden” bag rested in plain view on front seat of car equidistant from appellants and appellants were involved in ongoing criminal activity); In re F.T.J., 578 A.2d 1161 (D.C.1990) (per curiam) (concluding evidence sufficient to establish constructive possession of gun where appellant had been in car 15-20 minutes, there were three guns and three occupants in car, appellant had motive to possess gun, and manner in which gun protruded from under seat suggested it had been placed there from where appellant was seated); Tucker v. United States, 421 A.2d 32 (D.C.1980) (per curiam) (concluding evidence sufficient to support constructive possession where pistol protruded from under armrest in front seat next to where appellant had been sitting and appellant was seen in possession of the gun three days before arrest); Holley v. United States, 286 A.2d 222 (D.C.1972) (concluding evidence sufficient for constructive possession where gun was *661found on rear seat of car, only partially covered by a coat and within appellant’s reach from right front seat, and, when asked to leave vehicle, appellant lunged for coat). But see Hamilton v. United States, 395 A.2d 24 (D.C.1978) (concluding evidence sufficient to support finding of constructive possession where gun was found wedged between passenger seat and door and appellant had been sitting in passenger seat when vehicle was stopped); Kenhan v. United States, 263 A.2d 253 (D.C. 1970) (concluding evidence sufficient for constructive possession where butt of gun stuck out from between backrest and passenger seat to the left of where appellant had been sitting when the vehicle was stopped); Waterstaat v. United States, 252 A.2d 507 (D.C.1969) (concluding evidence sufficient to support constructive possession where gun was found on front seat between appellant driver and passenger).3
A review of our cases therefore shows that the holdings in Bernard and In re T.M., that intent to exercise dominion and control is a necessary element of constructive possession, is supported by most of our cases; but that the statements in Bur-nette and In re F.T.J., that the requisite element of intent may be satisfied by proof of presence of contraband in plain view in proximity to the defendant, appear to go beyond the facts of our decided cases on the issue, including the facts in Burnette and In re F.T.J. themselves. Thus, the cases that comprise our jurisprudence in this area suggest that the evidence in this case was less than we have deemed sufficient in the past for constructive possession.
There is serious doubt about the merits of permitting an inference — sufficient without more — of an individual’s intent to exercise dominion and control over contraband from the sole facts of that contraband’s presence in plain view and accessibility to an individual. The starting point must be the reason for the intent requirement. Here, unfortunately, our cases do not shed much light. Traditionally, constructive possession had been defined as: a) knowledge of the presence of contraband; and b) the ability to exercise dominion and control over the contraband. See Brown, swpra, 546 A.2d at 394; Tucker, supra, 421 A.2d at 35 (“To prove constructive possession, evidence must be adduced establishing that the [contraband] was conveniently accessible to appellant and that he knew of its presence.”) (citations omitted). In 1990, the Bernard and In re T.M. cases firmly established the intent element of the constructive possession test as part of our jurisprudence;4 however, *662neither case dealt with contraband located in a vehicle.5 In In re T.M., this court recognized the distinction, stating “[t]his is not a case such as Brown, supra, 546 A.2d at 396-97, or Waterstaat, supra, 252 A.2d at 509, in which it was held that the requisite inferences may be drawn from the location of weapons in plain view and substantially within a defendant’s reach in the closer confines of an automobile.” 577 A.2d at 1154 n. 12. In Brown and Waterst-aat, however, it is clear that the “requisite inferences” referred to were the traditional elements of knowledge and accessibility, rather than intent. See Brown, supra, 546 A.2d at 396 (holding that evidence of proximity is sufficient to permit jury to infer convenient access); Waterstaat, supra, 252 A.2d at 509 (inferring knowledge from fact that gun was in plain view and accessibility from trial court’s determination that gun was in close proximity to appellant). Nonetheless, in In re F.T.J., supra, the first case involving a vehicle following our explicit requirement of the intent prong in Bernard, this court extended the “requisite inferences” analysis to include intent. See 578 A.2d at 1162-63 (“[b]y ‘requisite inferences’ we meant not only access and knowledge but the intent to exercise dominion or control”) (emphasis added); see also Speight v. United States, 599 A.2d 794, 796 (D.C.1991) (citing In re T.M., supra, 577 A.2d at 1151 n. 5) (noting that to prove constructive possession, the government must show; “(1) that appellant knew of the location of the drugs; (2) that he had the ability to exercise dominion and control over them; and, (3) that he intended to guide their destiny”). While intent may have been an underlying element in the line of cases culminating in Bernard,6 the In re F.T.J. court applied the “requisite inferences” language to the recently recognized intent prong largely without discussion, based, apparently, on consideration of the “closer confines of an automobile.”7
Bearing in mind that the intent requirement was added to the traditional requirements of knowledge and accessibility of contraband, it is immediately questionable whether permitting an inference of intent from the same facts that establish knowledge and accessibility guts the imposition *663of the intent element. Further, assuming that intent is a requisite element for constructive possession, the question before us becomes whether permitting a sufficient inference of intent to possess contraband from presence in plain view and proximity of contraband is reasonable, rehable and consistent with the statutory offense.
With respect to the reasonableness of the inference, the first and obvious point is that we all know from personal experience that it is not necessarily so. Whenever we stand or sit in a bus, train, car or airplane, for example, we are of necessity in close proximity to the belongings of others and have an opportunity to see them and have ready access to them. We could, presumably, reach over to the seat next to ours, under the seat or into overhead bins and racks. That knowledge and proximity, however, do not mean that we intend to exercise control over our neighbor’s belongings — not even if we continue to have knowledge of those belongings and to be in such close proximity to them during a long flight or train, trip. Should the fact that the instant situation involves the passenger of a car make a difference? We have recognized that the “closet ] confines” of a car may allow certain inferences. In re F.T.J., supra, 578 A.2d at 1162. Trains and planes also are confined spaces, however, and, if we were to limit the rule to cars, what about the passenger of a taxi who finds contraband on the seat left by the former passenger? Is this a special rule for private vehicles? And if so, what is it about such vehicles that makes the inference a realistic one? Or does the inference follow from a presumed relationship among the occupants of a vehicle that may apply in certain situations (e .g., where the passengers have a prior relationship) but not to others {e.g., hitchhiker)? See County Court of Ulster County v. Allen, 442 U.S. 140, 163, 99 S.Ct. 2213, 60 L.Ed.2d 777 (1979) (holding that fact that respondents were not “ ‘hitchhikers or other casual passengers’ ” supported inference of gun possession).
In evaluating inferences, we must be conscious of a second point, that an inference, to be permissible, may not be based on another inference, or else the jury’s verdict risks coming too close to being speculative, not evidence-based. In cases inferring intent from knowledge and proximity, however, inferences upon inferences are rife. In Burnette, supra, 600 A.2d at 1084, we permitted an inference that the appellant knew, about, and had the ability to exercise dominion and control over, a weapon that he could feel where it lay at his feet under the floor mat, but refused to extend to that “plain feel” situation the inference of intent to exercise dominion and control that we had permitted in “plain view” cases. The opinion does not rely on the premise that an inference based on the sense of sight is more reliable than one based on the sense of touch, and it is difficult to see how such a hierarchy of sensory impressions could be justified in the abstract. For example, the aroma of marijuana is as good as, if not a better, indicator of the presence of the drug than its green weed-like appearance. See Minnick v. United States, 607 A.2d 519, 525 & n. 10 (D.C.1992) (holding that smell of marijuana alone is sufficient to establish probable cause to search vehicle). One possible reason for the Burnette court’s rejection of a “plain feel” predicate for an inference of intent might be the view that such a conclusion would require an inference based upon an inference. In Bur-nette we recognized that “the jury could properly infer that appellant must have placed his feet on the mat, felt a large hard object, lifted up the mat, and seen the gun.” 600 A.2d at 1083. Burnette held that this sufficed to establish the first two prongs of constructive possession: knowledge of the location of the gun and ability to exercise dominion and control over it. See id. The court was “unwilling,” however, to accept the “additional inference that he intended to exercise dominion over the gun — ‘to guide [its] destiny,’ ” Id. at 1084 (quoting In re T.M., supra, 577 A.2d at 1151 n. 5). The court did not explain its *664unwillingness. If the rationale was that the “plain feel” case involved one too many inferences, it is difficult to see how an inference of intent from “plain feel” is all that more speculative than the inference of knowledge and ability to control, permitted in Burnette, which were drawn from three cumulative inferences: an inference that appellant saw the gun drawn from an inference that he must have lifted the mat drawn from an inference that he must have placed his feet on the mat and felt the bulge. Perhaps the court felt that the “intent” requirement has a subjective component that should not be based exclusively on objective factors. However, that reluctance would apply with equal force to an inference of intent in plain view cases. The more likely reason, therefore, is that the court’s unwillingness in Burnette to extend the inference of intent to “plain feel” cases was due to the view that the “plain view” situation had already been decisively determined by the court, but that the rationale “should not be extended beyond [that] situation[ 8 Id. at 1084. In other words, the court refused to slip down the slope and engaged in classic line-drawing, albeit without explaining why the line was being drawn at that particular point. I agree with Burnette that it is prudent to restrict the inferences that may be drawn from other inferences, lest the jury be engaged in a possible — yet totally imagined — construct of the incident.
A third and, for me, decisive, reason for questioning the wisdom of inferring intent to possess from plain view and proximity of contraband is that by sanctioning such an inference in the course of an adjudication, we risk judicially altering the nature of the offense created by the legislature. If close proximity in a car to contraband in plain view is sufficient to permit an inference that conclusively establishes constructive possession, we are effectively imposing, as a matter of judicial interpretation, an obligation on an innocent passenger who discovers such contraband to take affirmative steps to disembark or in some other (at this point undefined) way, distance herself or himself from the offending substance. The statutory offense, however, is possession of a scheduled substance, not contiguous presence with a scheduled substance or failure to disassociate from such a substance. And how, realistically, is a passenger in a car to do so? By its very nature, a car is a confined space engaged in movement, which makes immediate distancing impractical and, possibly, even perilous. There is nothing in the language of D.C.Code § 33-541(a)(1) to imply that the legislature intended to impose such a duty on pain of conviction of possession. There are statutes that create a presumption that passengers in a car are deemed to possess contraband in the car, see Allen, supra, 442 U.S. at 142 n. 1, 99 S.Ct. 2213 (citing N.Y. Penal Law § 265.15(3) (McKinney 1967)), but our statute does not contain such a presumption.9 *665The dictionary meaning of “possess” implies a certain active dominion or control, “to have and hold as property ... be master of ... to take into one’s possession: seize or gain control of: make one’s own”. WEBSTER’S THIRD NEW INTERNATIONAL DICTIONARY 1770 (1986). It exceeds the judicial function to stray from the commonly-understood meaning of “possession” in defining “constructive possession” which, as we have recognized, is “a proxy for actual possession, e.g., in one’s hand, pocket or lap.” Burnette, supra, 600 A.2d at 1084. The inferences we sanction, as a judicially-created proxy, should not overtake the meaning that the legislature intended.
For these reasons I believe the full court could helpfully review the entire issue and decide whether to reaffirm that intent to exercise dominion and control is a required element of constructive possession, and, if so, whether plain view and proximity to contraband, standing alone, are sufficient to satisfy the intent requirement. Intent will almost always have to be inferred from other circumstances. Although, as Judge Farrell states in his concurrence, an objective starting-off point is useful, the desire for certainty should not overshadow fidelity to the statute’s purpose. Our case law confirms that even without the benefit of the inference, more probative evidence can be — and is — presented to support the element of intent. See cases cited, supra.
Turning to the case at hand, apart from the inference of intent from plain view and proximity, there was no evidence presented at trial that supported the conclusion that Rivas was connected to the drugs. The drugs were found by the police, who had been searching the car with a flashlight during a traffic stop that revealed an open container of alcohol leaning against the center of the back seat. Upon further investigation, the officers discovered two ziplock bags at about knee-level on the front console between the driver’s and the passenger’s seats, about one inch closer to the passenger’s seat. Following arrest, the officers discovered no drugs or money on Rivas. The police found $286 on Mel-gar, the driver, who testified that he had earned the money at his job. Before the car was stopped by the police, Rivas had left the car to speak to an individual on the sidewalk, leaving the front passenger side door open. The government claims that it did not rely solely on the inference of intent from proximity to the drugs in plain view, but also on Rivas’ “odd” and “elusive” behavior that suggested consciousness of guilt. According to the government, by leaving the car door open, Rivas indicated that he intended to return to the car, but did not do so and, instead, began to walk away when the police stopped the car. We have rejected that argument, however, in considering whether, other than by the inference, the government has presented evidence linking the accused to an ongoing criminal operation sufficient to supply evidence of the requisite intent. In Burnette we did not permit an inference of consciousness of guilt where the defendant “did not attempt to elude the police by driving away ..., there was no evidence of previous suspicious behavior on which the police were acting at the time they pulled *666over the car,10 ... there was no evidence about any of appellant’s activities earlier that night, ... and there is no evidence that the [contraband] in the front seat was in appellant’s plain view either earlier in the evening or at the time the police pulled over the jeep.” 600 A.2d at 1084. All of those elements of proof are also missing in this case. As in Burnette, there was no evidence here of the relationship between appellant and the driver. Moreover, in this case, even though Rivas had begun to walk away when the police stopped the car, he returned to the car when the police called to him. In this regard, this case is like In re M.I.W., 667 A.2d 573, 576 (D.C.1995), where we concluded that there was insufficient evidence to establish consciousness of guilt because we look for more than “walking away,” id. at 577, which may indicate merely an innocent desire to avoid contact with the police. See id. (citing Smith v. United States, 558 A.2d 312, 316 (D.C.1989) (en banc)). As in M.I.W., moreover, here “there is no evidence that [Rivas] had been in the car for a substantial period of time before it [was stopped], or that the vehicle had a functional interior light” that might have made it possible to see the drugs at night, at least when the car door was opened. Id. at 577. The testimony in this case, as in In re M.I.W., is that the police had to use a flashlight when they searched the car. In this case, in addition, the police testified that their search was aided by the dome light of their police car. Of course, the police car’s dome light was turned on after Rivas had exited the car to talk to someone on the sidewalk. As in here there was no evidence connecting Rivas to the car other than his being a passenger in it at the time, no evidence connecting Rivas to drug distribution and no evidence of drugs or money on his person, in contrast to the $236 found on Melgar, the driver.
Therefore, but for the categorical language in Burnette and In re F.T.J. that proximity to drugs in plain view permitted the jury to infer intent, I would conclude that the evidence was insufficient to prove Rivas’ intent to possess the drugs in question. However, as I am bound to follow these cases, I concur in affirming Rivas’ conviction.
. I also agree with the majority that the evidence was sufficient to convict Melgar, the driver and owner of the vehicle, and that the suppression motions were properly denied.
. At oral argument, counsel for Rivas represented that Rivas had just gotten into the car after he was offered a ride.
. Hamilton, Kenhan and Waterstaat were decided twelve to twenty-one years before we expressly set out the third "intent” prong for constructive possession in Bernard. See infra. The Hamilton court relied mainly on the element of accessibility to find constructive possession rather than the element of intent. Likewise, in Kenhan, the court addressed only the element of knowledge of the presence of contraband, not intent to possess it. And, in Waterstaat, the court focused on knowledge and accessibility, not intent. Moreover, in none of these cases was intent explicitly inferred from plain view and proximity.
. A number of other jurisdictions require intent as an element of constructive possession. See United States v. Armstrong, 1998 WL 278413, *3, 1998 U.S.App. LEXIS 10774, *3 (7th Cir.1998) (describing constructive possession as “ 'knowingly halving] the power and the intention at a given time to exercise dominion and control over an object, either directly or through others’ ”) (citation omitted); United States v. Hastings, 918 F.2d 369, 373 (2d Cir.1990) (defining constructive possession as " ‘knowingly having the power to and the intention at a given time to exercise dominion and control over an object, either directly or through others’ ”) (citation omitted); Palmer v. State, 593 So.2d 143, 145 (Ala.Crim.App.1991) (requiring proof of 1) actual or physical control; 2) intent to exercise dominion and control; and 3) external manifestations of intent and control over contraband) (citations omitted); McNulty v. State, 655 A.2d 1214, 1217 (Del.1995) (outlining elements of constructive possession as 1) knowledge of the contraband; 2) ability to exercise dominion and control over contraband; and 3) intent to guide its destiny); People v. McNeely, 99 Ill.App.3d 1021, 55 Ill.Dec. 321, 426 N.E.2d 296, 298 (1981) (noting that constructive possession is present when defendant has intent and capability to maintain *662control and possession of the contraband, and is aware of presence of contraband); Person v. State, 661 N.E.2d 587, 589 (Ind.Ct.App.1996) (finding constructive possession when defendant has intent and capability to maintain dominion and control over contraband); Commonwealth v. Spencer, 423 Pa.Super. 353, 621 A.2d 153, 155 (1993) (defining constructive possession as " 'the power to control the contraband and the intent to exercise that control’ ”) (citation omitted); State v. Reyes, 671 A.2d 1236, 1238 (R.I.1996) (noting that constructive possession is established by showing that accused knew of presence of contraband and had intent to exercise control over it); State v. Layman, 953 P.2d 782, 787 (Utah Ct.App.1998) (requiring nexus between accused and contraband sufficient to allow inference that accused had both ability and intent to exercise dominion and control over contraband).
. The Bernard case dealt with drugs found outdoors, while In re T.M. addressed guns located in an apartment.
. See Brown, supra, 546 A.2d at 394 n. 2 (referring to trial court's jury instruction describing constructive possession as "knowingly halving] both the power and the intention at a given time to exercise dominion and control over a thing”) (emphasis added); In re T.M., supra, 577 A.2d at 1151 n. 5 (recognizing that, traditionally, our case law has not explicitly specified that intent to guide destiny of contraband is an element of constructive possession, but suggesting that point may be explicit in certain decisions and noting explicit intent requirement in District of Columbia jury instruction on constructive possession).
.I have been unable to find another jurisdiction that, absent a statutory presumption, permits an inference of intent solely from plain view and proximity. See, e.g., Hishaw v. Oklahoma, 568 P.2d 643, 644-45 (Okla.Crim.App.1977) (finding evidence insufficient to prove possession where drugs were in plain view on floor of car near appellant's feet, and noting that "mere presence of the defendant in even such a confined area as an automobile interi- or in which illicit drugs are found does not, standing alone, constitute sufficient proof of his possession of such drugs”).
. As discussed earlier, however, the inferential Pandora's box had not necessarily been opened as wide as purported by the language in the "plain view" cases because the facts in those cases provided additional support for an inference of intent.
. Unlike in County Court of Ulster County v. Allen, supra, the issue before us is not the constitutionality of a presumption created by the legislature, but whether a judicially-created inference is consistent with the statutory offense of possession. In any event, it is worth noting that in that case, the Supreme Court did not pass on the facial constitutionality of New York’s statutory presumption that a person in a car that contains guns possesses the guns. Rather, the Supreme Court ruled, in an as-applied constitutional challenge to the statute, that the rebuttable statutory presumption was not unconstitutional because the facts in the case were sufficient to support a finding of constructive possession. 442 U.S. at 141, 99 S.Ct. 2213. In that case, four occupants of a car were convicted of possessing two loaded handguns that were found half secreted in the open handbag of a sixteen-year-old girl seated on the front passenger seat. The Court reasoned that the analysis was no different than if the guns had been in plain view and accessible to the other three occupants in the car. 442 U.S. at 163, 99 S.Ct. 2213. The Court emphasized an additional fact that linked the three occupants to the guns:
*665Moreover, it is highly improbable that the loaded guns belonged to Jane Doe [the 16-year-old girl] or that she was solely responsible for their being in her purse. As a 16-year-old girl in the company of three adult men she was the least likely of the four to be carrying one, let alone two, heavy handguns. It is far more probable that she relied on the pocketknife found in her brassiere for any necessary self-protection. Under these circumstances, it was not unreasonable for her counsel to argue and for the jury to infer that when the car was halted for speeding, the other passengers in the car anticipated the risk of a search and attempted to conceal their weapons in a pocketbook in the front seat. The inference is surely more likely than the notion that these weapons were the sole property of the 16-year-old girl.
Id. at 163-64, 99 S.Ct. 2213. Thus, the convictions were not based solely on the statutory presumption as the particular facts in the case provided additional evidence that all four occupants possessed the guns.
. In Burnette, the car was pulled over to investigate a traffic violation. Similarly, in this case, the police pulled over the car because it had broken down and obstructed the street for a few minutes. As soon as the police activated the cruiser’s lights, the car complied by moving to the curb.